Initial commit of person-detection-app
Change-Id: Ic4141fd19d3d8c89929b798191310fa2c49e3070
diff --git a/.gitignore b/.gitignore
new file mode 100644
index 0000000..483de3d
--- /dev/null
+++ b/.gitignore
@@ -0,0 +1,2 @@
+.venv
+.idea
diff --git a/LICENSES/LicenseRef-ONF-Member-1.0.txt b/LICENSES/LicenseRef-ONF-Member-1.0.txt
new file mode 100644
index 0000000..3625ce5
--- /dev/null
+++ b/LICENSES/LicenseRef-ONF-Member-1.0.txt
@@ -0,0 +1,295 @@
+NF Member-Only Software License
+Version 1.0
+June 2020
+
+This ONF Member-Only Software License (this "License") is entered into by and
+between Open Networking Foundation ("ONF") and you and/or the entity on whose
+behalf you are accessing the Works (as defined in Section 1 below) ("you").
+This License forms a legally binding contract between you and ONF in relation
+to your access to and use of the Works. You may not access or use the Works if
+you do not accept this License.
+
+By accessing or using the Works, you hereby agree to the terms of this License.
+You may not use the Works and may not accept this License if you are a person
+barred from receiving the Works under the laws of the United States or other
+countries, including the country from which you access or use the Works. If you
+are agreeing to be bound by this License on behalf of your employer or other
+entity, you represent and warrant that you have full legal authority to bind
+your employer or such entity to this License. If you do not have the requisite
+authority, you may not accept this License or access or use the Works on behalf
+of your employer or any other entity. You and ONF acknowledge and agree as
+follows:
+
+1. DEFINITIONS.
+
+"Source" form shall mean the preferred form for making modifications, including
+but not limited to software source code, documentation source, and
+configuration files.
+
+"Object" form shall mean any form resulting from mechanical transformation or
+translation of a Source form, including but not limited to compiled object
+code, generated documentation, and conversions to other media types.
+
+"Works" shall mean the works of authorship, whether in Source or Object form,
+made available under this License, as indicated by a copyright notice/header
+file that is included in or attached to the Works.
+
+2. BACKGROUND.
+
+The purpose of this License is to provide ONF member companies in good standing
+("Members") with early access to software projects currently in development by
+ONF prior to their public release, to allow Members to review and test the
+Works internally for the purpose of making further contributions to the
+applicable project. Members who are Partner-level Members at the time they
+first access the Works are additionally permitted under this License to use the
+Works internally for the purpose of building and developing products and
+services for commercial purposes as defined in Section 4. However, this
+License does not permit any Member, regardless of Member level, to redistribute
+all or any portion of the Works, except as part of a contribution back to the
+applicable project, in accordance with Section 8 below, or in binary form as
+provided for Partner-level Members in Section 4 below. You understand and
+agree that ONF has the sole right to sublicense and distribute a public version
+of the final code base for the applicable project to third parties under the
+2.0 version of the Apache License, or other permissive license terms selected
+by ONF in its sole discretion.
+
+The Purpose of this License is to incubate the development of new software
+projects by providing a framework for ONF Members to access and contribute to
+the projects while minimizing the potential for forking or misuse of the
+projects. Nothing in this Background paragraph, however, shall require ONF to
+take any particular course of action regarding any Works covered by this
+License.
+
+3. LIMITED LICENSE TO ALL MEMBERS IN GOOD STANDING.
+
+Subject to your complete and ongoing compliance with all the terms and
+conditions set forth in this License, including without limitation all license
+limitations and restrictions set forth herein, ONF grants you the following
+limited, non-exclusive, non-transferable, non-sublicensable, revocable license
+to use, and (where applicable) authorize your employees and contractors
+performing work on your behalf (but only if such contractors are themselves
+Members of ONF and accept the terms of this License) to use the Works
+internally solely in connection with (a) reviewing, testing, modifying and
+creating derivatives of the Works for the sole purpose of making contributions
+by you to the code repository for the applicable project maintained by ONF; and
+(b) building, developing and testing your own applications or components that
+may interoperate with or incorporate the Works ("Applications") for the
+purposes of internal evaluation only.
+
+4. ADDITIONAL LICENSE TO PARTNER-LEVEL MEMBERS IN GOOD STANDING.
+
+Each project subject to this License shall be declared by ONF to be in one of
+two phases: a) Incubation Phase, or b) Partner Commercialization Phase. Each
+project shall start in the Incubation Phase, and ONF, at its sole discretion,
+shall determine when each project moves into the Partner Commercialization
+Phase. Each project shall be documented to be in one of these two phases in
+the Project's README documentation and can be confirmed by contacting ONF.
+
+Once any project moves into the Partner Commercialization Phase, if you are a
+Partner-level ONF Member in good standing at the time you first access the
+Works and for the entire period during which you access and use the Works, in
+addition to the rights granted in Section 3, and subject to your complete and
+ongoing compliance with all the terms and conditions set forth in this License,
+including without limitation all license limitations and restrictions set forth
+herein, ONF grants you the limited, non-exclusive, non-transferable,
+non-sublicensable, revocable license to use, and (where applicable) authorize
+your employees and contractors performing work on your behalf (but only if such
+contractors are themselves Members of ONF and accept the terms of this License)
+to use, test, modify and create derivatives of the Works internally for the
+purpose of building, developing, testing, marketing, offering for sale and
+selling commercial products and services to third parties. Wherever possible,
+you will take reasonable precautions to ensure Works remain accessible only to
+ONF Members, and any commercial product or service requiring redistribution of
+the Works, of modifications of the Works or of derivatives of the Works must be
+made solely in binary machine code form where possible to further this aim.
+
+Partners-level Members gaining additional insights as a result of accessing and
+using the Works for commercial products and services are expected to provide
+feedback to and to make contributions to the project for the benefit of both
+the project and the general membership.
+
+You may verify your company's Partner-level membership status on the ONF
+website at: https://www.opennetworking.org/member-listing/.
+
+4.1 GRACE PERIOD FOR NON-PARTNER-LEVEL MEMBERS
+
+If you are not a Partner-level Member and you wish to benefit from the rights
+granted to ONF Partners in Section 4, you have a 90-day grace period from the
+day you first access the Works to upgrade to a Partner-Level ONF Membership.
+Should you upgrade within this grace period, you will be recognized as if you
+had been a Partner on the day you first accessed the Works for the purpose of
+interpreting your rights under Section 4.
+
+5. RESTRICTIONS.
+
+By accessing or using the Works, you represent, warrant, and covenant that you
+and/or the entity on whose behalf you are accessing the Works is a Member in
+good standing of ONF.
+
+You acknowledge that, except for the limited rights expressly set forth above,
+the foregoing licenses do not include any right to, and you agree not to (i)
+redistribute, sell, lease, license, modify or otherwise create any derivative
+works of any portion of the Works, or (ii) distribute, deploy, or otherwise
+utilize the Works or Applications for any public, production, commercial, or
+other similar purpose other than internal use for evaluation and development of
+contributions to the code repository for the applicable project maintained by
+ONF and non-public, experimental Applications.
+
+Except as expressly provided herein, you may not (w) reproduce, distribute,
+publicly display, or publicly perform any part of the Works, (x) decompile,
+reverse engineer, or otherwise access or attempt to access source code not made
+available to you in source code form, (y) make or attempt to make any
+modification to, or otherwise create any derivative works of, the Works; or (z)
+remove, obscure, interfere with or circumvent any feature of the Works,
+including without limitation any copyright or other intellectual property
+notices, security, or access control mechanism. You agree that you will not use
+the Works to do anything illegal, including facilitating, promoting, or
+otherwise encouraging any illegal activities.
+
+You represent and warrant that the Works will not be shipped, transferred or
+exported into any country or used in any manner prohibited by the United States
+Export Administration Act or any other export laws, restrictions or regulations
+(collectively the "Export Laws"). In addition, if the Works are identified as
+an export controlled item under the Export Laws, you represent and warrant that
+you are not a citizen, or otherwise located within, an embargoed nation
+(including without limitation Crimea, Cuba, Iran, North Korea, Sudan, or Syria)
+and that you are not otherwise prohibited under the Export Laws from receiving
+the Works.
+
+ANY USE IN VIOLATION OF THE FOREGOING LIMITATIONS AND RESTRICTIONS IS STRICTLY
+PROHIBITED, AND UNLICENSED.
+
+6. RESERVATION OF RIGHTS.
+
+The Works are owned by ONF and/or its licensors, and are licensed, not sold, to
+you. The Works are protected by copyright, trade dress, patent, and trademark
+laws of the United States and other jurisdictions, international conventions,
+and all other relevant intellectual property and proprietary rights, and
+applicable laws (collectively, the "Intellectual Property Rights"). You agree
+that ONF or its subsidiaries or affiliated companies and/or its licensors own
+all legal right, title and interest in and to the Works, including any and all
+Intellectual Property Rights. ONF reserves all rights not expressly granted in
+this License. You do not acquire any right, title or interest to the Works,
+whether by implication, estoppel, or otherwise, except for the limited rights
+set forth in this License.
+
+You agree that the form and nature of the Works that ONF provides may change
+without prior notice to you and that future versions of the Works may be
+incompatible with applications developed using previous versions of the Works.
+You agree that ONF may stop (permanently or temporarily) providing the Works
+(or any features within the Works) to you or to users generally at ONF's sole
+discretion, without prior notice to you. Nothing in this License gives you a
+right to use any of ONF's trade names, trademarks, service marks, logos, domain
+names, or other distinctive brand features.
+
+7. PURPOSE AND USE.
+
+You agree that you will not engage in any activity with the Works, including
+the development or distribution of an Application that interferes with,
+disrupts, damages, or accesses in an unauthorized manner the servers, networks,
+or other properties or services of any third party. You agree that you are
+solely responsible for (and that ONF has no responsibility to you or to any
+third party for) any data, content, or resources that you create, transmit or
+display. You agree that you are solely responsible for (and that ONF has no
+responsibility to you or to any third party for) any breach of your obligations
+under this License, any applicable third party contract or terms of service, or
+any applicable law or regulation, and for the consequences (including any loss
+or damage which ONF or any third party may suffer) of any such breach.
+
+8. FEEDBACK; CONTRIBUTIONS.
+
+If you provide ONF with any comments, bug reports, feedback, enhancements, or
+modifications proposed or suggested by you for the Works ("Feedback"), such
+Feedback will be submitted as a Contribution to the applicable project
+maintained by ONF under the terms and conditions of a Contributor License
+Agreement between you and ONF, is provided on a non-confidential basis
+(notwithstanding any notice to the contrary you may include in any accompanying
+communication), and ONF shall have the right to use such Feedback at its
+discretion, including, but not limited to the incorporation of such Feedback
+into the Works and any final code base that may be derived from the Works.
+
+9. TERMINATION OF THIS LICENSE.
+
+This License will continue to apply until terminated by either you or ONF as
+set out below. If you want to terminate this License, you may do so by ceasing
+your use of the Works. If you cease being a Member in good standing of ONF,
+this License will terminate automatically and without further notice to you.
+ONF may at any time terminate this License: (a) if you have breached any
+provision of this License; or (b) at ONF's convenience. When this License
+comes to an end, except as otherwise set forth in this paragraph 9, all of the
+legal rights, obligations and liabilities that you and ONF have benefited from,
+been subject to (or which have accrued over time while this License has been in
+force) or which are expressed to continue indefinitely, shall be unaffected by
+this cessation, and paragraphs 1, 2 and 5 through 14 shall continue to apply to
+such rights, obligations and liabilities indefinitely. In addition, unless
+this License is terminated by ONF as a result of your breach of any provision
+of this License, Partner-level Members entitled to the benefits of Section 4 of
+this License shall also retain all rights granted by Section 4 so as long as
+you remain a Partner-level Member in good standing.
+
+10. DISCLAIMER OF WARRANTIES.
+
+YOU EXPRESSLY UNDERSTAND AND AGREE THAT YOUR USE OF THE WORKS IS AT YOUR SOLE
+RISK AND THAT THE WORKS ARE PROVIDED "AS IS" AND "AS AVAILABLE" WITHOUT
+WARRANTY OF ANY KIND FROM ONF. YOUR USE OF THE WORKS IS AT YOUR OWN DISCRETION
+AND RISK AND YOU ARE SOLELY RESPONSIBLE FOR ANY DAMAGE TO YOUR COMPUTER SYSTEM
+OR OTHER DEVICE OR LOSS OF DATA THAT RESULTS FROM SUCH USE. ONF FURTHER
+EXPRESSLY DISCLAIMS ALL WARRANTIES AND CONDITIONS OF ANY KIND, WHETHER EXPRESS
+OR IMPLIED, INCLUDING, BUT NOT LIMITED TO THE IMPLIED WARRANTIES AND CONDITIONS
+OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. ONF
+EXPRESSLY DISCLAIMS ANY WARRANTIES OF ANY KIND WITH RESPECT TO THE ACCURACY OR
+FUNCTIONALITY OF THE WORKS, AND WITH RESPECT TO THE ACCURACY, VALIDITY, OR
+COMPLETENESS OF ANY INFORMATION OR FEATURES AVAILABLE THROUGH THE WORKS, OR THE
+QUALITY OR CONSISTENCY OF THE WORKS OR RESULTS OBTAINED THROUGH THEIR USE.
+
+11. LIMITATION OF LIABILITY.
+
+YOU EXPRESSLY UNDERSTAND AND AGREE THAT ONF, ITS SUBSIDIARIES AND AFFILIATES,
+AND ITS LICENSORS SHALL NOT BE LIABLE TO YOU UNDER ANY THEORY OF LIABILITY FOR
+ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, CONSEQUENTIAL OR EXEMPLARY DAMAGES
+THAT MAY BE INCURRED BY YOU, INCLUDING ANY LOSS OF DATA, WHETHER OR NOT ONF OR
+ITS REPRESENTATIVES HAVE BEEN ADVISED OF OR SHOULD HAVE BEEN AWARE OF THE
+POSSIBILITY OF ANY SUCH LOSSES. IN NO EVENT WILL ONF'S AGGREGATE LIABILITY FOR
+DAMAGES ARISING OUT OF THIS LICENSE OR YOUR USE OF THE WORKS EXCEED THE GREATER
+OF AMOUNTS PAID BY YOU FOR THE WORKS, IF ANY.
+
+12. INDEMNIFICATION.
+
+To the maximum extent permitted by law, you agree to defend, indemnify and hold
+harmless ONF, its affiliates and their respective directors, officers,
+employees and agents from and against any and all claims, actions, suits or
+proceedings, as well as any and all losses, liabilities, damages, costs and
+expenses (including reasonable attorneys' fees) arising out of or accruing from
+(a) your use of the Works, (b) any Application or (for Partner-level Members)
+commercial product or service you develop using the Works that infringes any
+copyright, trademark, trade secret, trade dress, patent or other intellectual
+property right of any person, and (c) any non-compliance by you with this
+License or any applicable law or regulation.
+
+13. CHANGES TO THIS LICENSE.
+
+ONF may make changes to this License as it distributes new versions of the
+Works. When these changes are made, ONF will make a new version of this License
+available on the ONF website.
+
+14. MISCELLANEOUS TERMS.
+
+This License constitutes the entire legal agreement between you and ONF with
+regard to your access and use of the Works, and governs your access and use of
+the Works and completely supersedes any prior agreements between you and ONF in
+relation to the Works, except for any Contributor License Agreement you may
+have executed with ONF. You agree that if ONF does not exercise or enforce any
+legal right or remedy which is contained in this License (or which ONF has the
+benefit of under any applicable law), this will not be taken to be a formal
+waiver of ONF's rights and that those rights or remedies will still be
+available to ONF. If any court of law, having the jurisdiction to decide on
+this matter, rules that any provision of this License is invalid, then that
+provision will be removed from this License without affecting the rest of this
+License. The remaining provisions of this License will continue to be valid and
+enforceable so long as this License, as so modified, does not substantially
+impair the respective expectations or reciprocal obligations of the parties.
+The rights granted in this License may not be assigned or transferred by you
+without the prior written approval of ONF. You shall not be permitted to
+delegate your responsibilities or obligations under this License without the
+Prior written approval of ONF.
+
diff --git a/Makefile b/Makefile
index e057470..e4ca58d 100644
--- a/Makefile
+++ b/Makefile
@@ -1,5 +1,67 @@
-all:
+# SPDX-FileCopyrightText: 2020-present Open Networking Foundation <info@opennetworking.org>
+# SPDX-License-Identifier: LicenseRef-ONF-Member-1.01
+#
+define PROJECT_HELP_MSG
+Usage:
+ make help show this message
+ make clean remove intermediate files
-build:
+ make ${VENV} make a virtualenv in the base directory
+ make python-reqs install python packages in requirements.pip
+ make git-config set local git configuration
+ make setup make python-reqs
+
+ make run launch network-video-recorder
+ make run-native run native application (no docker)
+ make run-native-no-show run native applicaiton (no docker) w/o video output
+endef
+export PROJECT_HELP_MSG
+
+SHELL = /bin/bash
+VENV = .venv
+PYTHON = $(VENV)/bin/python3
+PIP = $(VENV)/bin/pip
+IMAGE = person-detection-app
+
+help:
+ echo "$$PROJECT_HELP_MSG" | less
+
+
+$(VENV): $(VENV)/touchfile
+
+$(VENV)/touchfile: requirements.txt
+ test -d $(VENV) || python3 -m venv $(VENV)
+ . $(VENV)/bin/activate; pip install -Ur requirements.txt
+ touch $(VENV)/touchfile
+
+$(VENV)/bin/activate: requirements.txt
+ python3 -m venv $(VENV)
+ $(PIP) install -r requirements.txt
+
+build: $(VENV)
+ docker build -f docker/Dockerfile -t $(IMAGE) .
+
+run:
+ docker run -itu root:root --privileged --network host --name $(IMAGE) --rm $(IMAGE)
+
+run-native-test:
+ . ./person_detection.sh -i ./resources/run.mp4
+
+run-native:
+ . ./person_detection.sh -i gstreamer
+
+run-native-test-no-show:
+ . ./person_detection.sh -i ./resources/run.mp4 -ns
+
+run-native-no-show:
+ . ./person_detection.sh -i gstreamer -ns
test:
+
+
+CLEANUP = *.pyc $(VENV)
+clean:
+ rm -rf ${CLEANUP}
+
+.PHONY: run clean
+
diff --git a/README.md b/README.md
index 8485038..94e9c2a 100644
--- a/README.md
+++ b/README.md
@@ -1,3 +1,25 @@
# Persion-detection Applicaiton
person-detection is a demo application that runs in the Aether edge and showcases Aether's support for for low-latency, AI/ML driven IOT applications that make use of Aether's support for end-to-end slicing.
+
+## Build base image
+```
+git clone git@github.com:openvinotoolkit/docker_ci.git
+cd docker_ci
+python3 docker_openvino.py build -d cpu -os ubuntu18 --distribution runtime --product_version 2021.4
+```
+
+## Build person-detection
+```
+make build
+```
+
+## Start RPi camera
+```
+gst-launch-1.0 rpicamsrc bitrate=1000000 ! video/x-raw,width=640,height=480,framerate=25/1,profile=baseline ! jpegenc ! rtpjpegpay ! udpsink host=10.128.99.49 port=5000
+```
+
+## Start persion-detection
+```
+make run
+```
diff --git a/docker/Dockerfile b/docker/Dockerfile
new file mode 100644
index 0000000..6fcc485
--- /dev/null
+++ b/docker/Dockerfile
@@ -0,0 +1,27 @@
+#python3 docker_openvino.py build -d cpu -os ubuntu18 --distribution runtime --product_version 2021.4
+FROM ubuntu18_runtime:2021.4
+
+USER root
+
+RUN apt-get update && apt-get install -y \
+ wget \
+ unzip \
+ libglib2.0-0 \
+ libsm6 \
+ libxrender1 \
+ libxext6 \
+ vim \
+ ffmpeg \
+ x11-apps
+
+RUN apt-get install -y libgstreamer1.0-dev libgstreamer-plugins-base1.0-dev libgstreamer-plugins-bad1.0-dev gstreamer1.0-plugins-base gstreamer1.0-plugins-good gstreamer1.0-plugins-bad gstreamer1.0-plugins-ugly gstreamer1.0-libav gstreamer1.0-doc gstreamer1.0-tools gstreamer1.0-x gstreamer1.0-alsa gstreamer1.0-gl gstreamer1.0-gtk3 gstreamer1.0-qt5 gstreamer1.0-pulseaudio
+
+WORKDIR /var/person-detection-app
+
+COPY . .
+
+RUN . .venv/bin/activate && \
+ pip3 install -r requirements.txt
+
+ENTRYPOINT ["./person_detection.sh"]
+CMD ["-i gstreamer", "-ns"]
diff --git a/person_detection.sh b/person_detection.sh
new file mode 100755
index 0000000..a58351c
--- /dev/null
+++ b/person_detection.sh
@@ -0,0 +1,5 @@
+#!/bin/bash
+args="$@"
+source .venv/bin/activate
+source /opt/intel/openvino/bin/setupvars.sh
+python3 person_detection/person_detection.py -d CPU -m ./resources/models/intel/person-detection-retail-0013/FP32/person-detection-retail-0013.xml $args
diff --git a/person_detection/person_detection.py b/person_detection/person_detection.py
new file mode 100644
index 0000000..037734c
--- /dev/null
+++ b/person_detection/person_detection.py
@@ -0,0 +1,273 @@
+"""
+SPDX-FileCopyrightText: 2020-present Open Networking Foundation <info@opennetworking.org>
+SPDX-License-Identifier: LicenseRef-ONF-Member-1.01
+"""
+
+from __future__ import print_function
+
+import logging as log
+import os
+import sys
+import time
+from argparse import ArgumentParser, SUPPRESS
+
+import cv2
+from imutils import build_montages
+from openvino.inference_engine import IECore
+
+
+def build_argparser():
+ parser = ArgumentParser(add_help=False)
+ args = parser.add_argument_group('Options')
+ args.add_argument('-h', '--help', action='help', default=SUPPRESS, help='Show this help message and exit.')
+ args.add_argument("-m", "--model", help="Required. Path to an .xml file with a trained model.",
+ required=True, type=str)
+ args.add_argument("-i", "--input",
+ help="Required. Path to video file or image. 'cam' for capturing video stream from camera",
+ required=True, type=str)
+ # args.add_argument("-i2", "--input2",
+ # help="Optional. Path to second video file or image. 'cam' for capturing video stream from camera",
+ # default=None, type=str)
+ args.add_argument("-l", "--cpu_extension",
+ help="Optional. Required for CPU custom layers. Absolute path to a shared library with the "
+ "kernels implementations.", type=str, default=None)
+ args.add_argument("-pp", "--plugin_dir", help="Optional. Path to a plugin folder", type=str, default=None)
+ args.add_argument("-d", "--device",
+ help="Optional. Specify the target device to infer on; CPU, GPU, FPGA, HDDL or MYRIAD is "
+ "acceptable. The demo will look for a suitable plugin for device specified. "
+ "Default value is CPU", default="CPU", type=str)
+ args.add_argument("--labels", help="Optional. Path to labels mapping file", default=None, type=str)
+ args.add_argument("-pt", "--prob_threshold", help="Optional. Probability threshold for detections filtering",
+ default=0.5, type=float)
+ args.add_argument("-ns", help='No show output', action='store_true')
+
+ return parser
+
+
+def main():
+ log.basicConfig(format="[ %(levelname)s ] %(message)s", level=log.INFO, stream=sys.stdout)
+ args = build_argparser().parse_args()
+ model_xml = args.model
+ model_bin = os.path.splitext(model_xml)[0] + ".bin"
+ # Plugin initialization for specified device and load extensions library if specified
+ log.info("Initializing plugin for {} device...".format(args.device))
+ # plugin = IEPlugin(device=args.device, plugin_dirs=args.plugin_dir)
+ # if args.cpu_extension and 'CPU' in args.device:
+ # plugin.add_cpu_extension(args.cpu_extension)
+ # Read IR
+ log.info("Reading IR...")
+ net = IECore().read_network(model=model_xml, weights=model_bin)
+
+ assert len(net.inputs.keys()) == 1, "Demo supports only single input topologies"
+ assert len(net.outputs) == 1, "Demo supports only single output topologies"
+ input_blob = next(iter(net.inputs))
+ out_blob = next(iter(net.outputs))
+
+ # input_blob2 = next(iter(net.inputs))
+ # out_blob2 = next(iter(net.outputs))
+
+ log.info("Loading IR to the plugin...")
+ # exec_net = IECore().load_network(network=net, device_name=args.device, num_requests=2)
+ exec_net = IECore().load_network(network=net, device_name=args.device, num_requests=1)
+ # Read and pre-process input image
+ n, c, h, w = net.inputs[input_blob].shape
+ # n2, c2, h2, w2 = net.inputs[input_blob2].shape
+ del net
+ if args.input == 'cam':
+ input_stream = 0
+ elif args.input == 'gstreamer':
+ # gst rtp sink
+ input_stream = 'udpsrc port=5000 caps = " application/x-rtp, encoding-name=JPEG,payload=26" ! rtpjpegdepay ! decodebin ! videoconvert ! appsink'
+ #input_stream = 'udpsrc port=5000 caps = "application/x-rtp, media=(string)video, clock-rate=(int)90000, encoding-name=(string)H264, payload=(int)96" ! rtph264depay ! decodebin ! videoconvert ! appsink'
+ else:
+ input_stream = args.input
+ assert os.path.isfile(args.input), "Specified input file doesn't exist"
+
+ if input_stream == 'gstreamer':
+ cap = cv2.VideoCapture(input_stream, cv2.CAP_GSTREAMER)
+ else:
+ cap = cv2.VideoCapture(input_stream)
+
+ # if args.input2 == 'cam':
+ # input_stream2 = 0
+ # elif args.input2 == 'gstreamer':
+ # input_stream2 = 'udpsrc port=5001 caps = " application/x-rtp, encoding-name=JPEG,payload=26" ! rtpjpegdepay ! decodebin ! videoconvert ! appsink'
+ # else:
+ # input_stream2 = args.input2
+ # assert os.path.isfile(args.input2), "Specified input file doesn't exist"
+ if args.labels:
+ with open(args.labels, 'r') as f:
+ labels_map = [x.strip() for x in f]
+ else:
+ labels_map = None
+
+ # if input_stream2 == 'gstreamer':
+ # cap2 = cv2.VideoCapture(input_stream2, cv2.CAP_GSTREAMER)
+ # else:
+ # cap2 = cv2.VideoCapture(input_stream2)
+
+ cur_request_id = 0
+ next_request_id = 1
+
+ # cur_request_id2 = 1
+ # next_request_id2 = 0
+
+ log.info("Starting inference in async mode...")
+ log.info("To switch between sync and async modes press Tab button")
+ log.info("To stop the demo execution press Esc button")
+
+ # Async doesn't work if True
+ # Request issues = Runtime Error: [REQUEST BUSY]
+ is_async_mode = False
+ render_time = 0
+ ret, frame = cap.read()
+ # ret2, frame2 = cap2.read()
+
+ # Montage width and height
+ # In this case means 2x1 boxes
+ mW = 2
+ mH = 1
+
+ frameList = []
+
+ print("To close the application, press 'CTRL+C' or any key with focus on the output window")
+ # while cap.isOpened() or cap2.isOpened():
+ while cap.isOpened():
+ if is_async_mode:
+ ret, next_frame = cap.read()
+ # ret2, next_frame2 = cap2.read()
+ else:
+ ret, frame = cap.read()
+ # ret2, frame2 = cap2.read()
+ #if not (ret and ret2):
+ if not ret:
+ break
+ initial_w = cap.get(3)
+ initial_h = cap.get(4)
+ # initial_w2 = cap2.get(3)
+ # initial_h2 = cap2.get(4)
+ # Main sync point:
+ # in the truly Async mode we start the NEXT infer request, while waiting for the CURRENT to complete
+ # in the regular mode we start the CURRENT request and immediately wait for it's completion
+ inf_start = time.time()
+ if is_async_mode:
+ # if ret and ret2:
+ if ret:
+ in_frame = cv2.resize(next_frame, (w, h))
+ in_frame = in_frame.transpose((2, 0, 1)) # Change data layout from HWC to CHW
+ in_frame = in_frame.reshape((n, c, h, w))
+ exec_net.start_async(request_id=next_request_id, inputs={input_blob: in_frame})
+
+ # in_frame2 = cv2.resize(next_frame2, (w2, h2))
+ # in_frame2 = in_frame2.transpose((2, 0, 1)) # Change data layout from HWC to CHW
+ # in_frame2 = in_frame2.reshape((n2, c2, h2, w2))
+ # exec_net.start_async(request_id=next_request_id2, inputs={input_blob2: in_frame2})
+
+ else:
+ # if (ret and ret2):
+ if ret:
+ in_frame = cv2.resize(frame, (w, h))
+ in_frame = in_frame.transpose((2, 0, 1)) # Change data layout from HWC to CHW
+ in_frame = in_frame.reshape((n, c, h, w))
+ exec_net.start_async(request_id=cur_request_id, inputs={input_blob: in_frame})
+
+ # in_frame2 = cv2.resize(frame2, (w2, h2))
+ # in_frame2 = in_frame2.transpose((2, 0, 1)) # Change data layout from HWC to CHW
+ # in_frame2 = in_frame2.reshape((n2, c2, h2, w2))
+ # exec_net.start_async(request_id=cur_request_id2, inputs={input_blob2: in_frame2})
+
+ # if exec_net.requests[cur_request_id].wait(-1) == 0 and exec_net.requests[cur_request_id2].wait(-1) == 0:
+ if exec_net.requests[cur_request_id].wait(-1) == 0:
+ inf_end = time.time()
+ det_time = inf_end - inf_start
+
+ # Parse detection results of the current request
+ res = exec_net.requests[cur_request_id].outputs[out_blob]
+ # res2 = exec_net.requests[cur_request_id2].outputs[out_blob2]
+
+ for obj in res[0][0]:
+ # Draw only objects when probability more than specified threshold
+ if obj[2] > args.prob_threshold:
+ xmin = int(obj[3] * initial_w)
+ ymin = int(obj[4] * initial_h)
+ xmax = int(obj[5] * initial_w)
+ ymax = int(obj[6] * initial_h)
+ class_id = int(obj[1])
+ # Draw box and label\class_id
+ color = (min(class_id * 12.5, 255), min(class_id * 7, 255), min(class_id * 5, 255))
+ cv2.rectangle(frame, (xmin, ymin), (xmax, ymax), color, 2)
+ det_label = labels_map[class_id] if labels_map else str(class_id)
+ cv2.putText(frame, det_label + ' ' + str(round(obj[2] * 100, 1)) + ' %', (xmin, ymin - 7),
+ cv2.FONT_HERSHEY_COMPLEX, 0.6, color, 1)
+ print('Object detected, class_id:', class_id, 'probability:', obj[2], 'xmin:', xmin, 'ymin:', ymin,
+ 'xmax:', xmax, 'ymax:', ymax)
+
+ # for obj in res2[0][0]:
+ # # Draw only objects when probability more than specified threshold
+ # if obj[2] > args.prob_threshold:
+ # xmin = int(obj[3] * initial_w2)
+ # ymin = int(obj[4] * initial_h2)
+ # xmax = int(obj[5] * initial_w2)
+ # ymax = int(obj[6] * initial_h2)
+ # class_id = int(obj[1])
+ # # Draw box and label\class_id
+ # color = (min(class_id * 12.5, 255), min(class_id * 7, 255), min(class_id * 5, 255))
+ # cv2.rectangle(frame2, (xmin, ymin), (xmax, ymax), color, 2)
+ # det_label = labels_map[class_id] if labels_map else str(class_id)
+ # cv2.putText(frame2, det_label + ' ' + str(round(obj[2] * 100, 1)) + ' %', (xmin, ymin - 7),
+ # cv2.FONT_HERSHEY_COMPLEX, 0.6, color, 1)
+ # print('Object detected, class_id:', class_id, 'probability:', obj[2], 'xmin:', xmin, 'ymin:', ymin,
+ # 'xmax:', xmax, 'ymax:', ymax)
+
+ # Draw performance stats
+ inf_time_message = "Inference time: Not applicable for async mode" if is_async_mode else \
+ "Inference time: {:.3f} ms".format(det_time * 1000)
+ render_time_message = "OpenCV rendering time: {:.3f} ms".format(render_time * 1000)
+ if is_async_mode:
+ async_mode_message = "Async mode is on. Processing request {}".format(cur_request_id)
+ else:
+ async_mode_message = "Async mode is off. Processing request {}".format(cur_request_id)
+
+ cv2.putText(frame, inf_time_message, (15, 15), cv2.FONT_HERSHEY_COMPLEX, 0.5, (200, 10, 10), 1)
+ cv2.putText(frame, render_time_message, (15, 30), cv2.FONT_HERSHEY_COMPLEX, 0.5, (10, 10, 200), 1)
+ cv2.putText(frame, async_mode_message, (10, int(initial_h - 20)), cv2.FONT_HERSHEY_COMPLEX, 0.5,
+ (10, 10, 200), 1)
+
+ # cv2.putText(frame2, inf_time_message, (15, 15), cv2.FONT_HERSHEY_COMPLEX, 0.5, (200, 10, 10), 1)
+ # cv2.putText(frame2, render_time_message, (15, 30), cv2.FONT_HERSHEY_COMPLEX, 0.5, (10, 10, 200), 1)
+ # cv2.putText(frame2, async_mode_message, (10, int(initial_h - 20)), cv2.FONT_HERSHEY_COMPLEX, 0.5,
+ # (10, 10, 200), 1)
+
+ render_start = time.time()
+
+ if not args.ns:
+ # if ret and ret2:
+ if ret:
+ # frameList.append(frame)
+ # # frameList.append(frame2)
+ # montages = build_montages(frameList, (640, 480), (mW, mH))
+ # for montage in montages:
+ # cv2.imshow("Detection results", montage)
+ cv2.imshow("Detection results", frame)
+ render_end = time.time()
+ render_time = render_end - render_start
+
+ if is_async_mode:
+ cur_request_id, next_request_id = next_request_id, cur_request_id
+
+ frame = next_frame
+ # frame2 = next_frame2
+ key = cv2.waitKey(1)
+ if key == 27:
+ break
+ if 9 == key:
+ is_async_mode = not is_async_mode
+ log.info("Switched to {} mode".format("async" if is_async_mode else "sync"))
+
+ cap.release()
+ # cap2.release()
+ cv2.destroyAllWindows()
+
+
+if __name__ == '__main__':
+ sys.exit(main() or 0)
diff --git a/requirements.txt b/requirements.txt
new file mode 100644
index 0000000..9d6d5da
--- /dev/null
+++ b/requirements.txt
@@ -0,0 +1,2 @@
+imutils==0.5.4
+numpy==1.19.5
diff --git a/resources/models/intel/person-detection-retail-0013/FP16-INT8/person-detection-retail-0013.bin b/resources/models/intel/person-detection-retail-0013/FP16-INT8/person-detection-retail-0013.bin
new file mode 100644
index 0000000..dec950a
--- /dev/null
+++ b/resources/models/intel/person-detection-retail-0013/FP16-INT8/person-detection-retail-0013.bin
Binary files differ
diff --git a/resources/models/intel/person-detection-retail-0013/FP16-INT8/person-detection-retail-0013.xml b/resources/models/intel/person-detection-retail-0013/FP16-INT8/person-detection-retail-0013.xml
new file mode 100644
index 0000000..7bf3027
--- /dev/null
+++ b/resources/models/intel/person-detection-retail-0013/FP16-INT8/person-detection-retail-0013.xml
@@ -0,0 +1,35067 @@
+<?xml version="1.0" ?>
+<net name="ResMobNet_v4 (LReLU) with single SSD head" version="10">
+ <layers>
+ <layer id="0" name="3284328819830" type="Const" version="opset1">
+ <data element_type="f16" offset="0" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="1" name="3285328922773" type="Const" version="opset1">
+ <data element_type="f16" offset="2" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="2" name="3286329020898" type="Const" version="opset1">
+ <data element_type="f16" offset="0" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="3" name="3287329120439" type="Const" version="opset1">
+ <data element_type="f16" offset="2" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="4" name="4284428822554" type="Const" version="opset1">
+ <data element_type="f16" offset="4" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="5" name="4285428922101" type="Const" version="opset1">
+ <data element_type="f16" offset="6" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="6" name="4286429021780" type="Const" version="opset1">
+ <data element_type="f16" offset="4" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="7" name="4287429120886" type="Const" version="opset1">
+ <data element_type="f16" offset="6" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="8" name="5444544820604" type="Const" version="opset1">
+ <data element_type="f16" offset="8" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="9" name="5445544919815" type="Const" version="opset1">
+ <data element_type="f16" offset="10" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="10" name="5446545020175" type="Const" version="opset1">
+ <data element_type="f16" offset="8" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="11" name="5447545122041" type="Const" version="opset1">
+ <data element_type="f16" offset="10" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="12" name="2964296819803" type="Const" version="opset1">
+ <data element_type="f16" offset="12" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="13" name="2965296922185" type="Const" version="opset1">
+ <data element_type="f16" offset="14" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="14" name="2966297020562" type="Const" version="opset1">
+ <data element_type="f16" offset="12" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="15" name="2967297122368" type="Const" version="opset1">
+ <data element_type="f16" offset="14" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="16" name="5084508822452" type="Const" version="opset1">
+ <data element_type="f16" offset="16" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="17" name="5085508920694" type="Const" version="opset1">
+ <data element_type="f16" offset="18" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="18" name="5086509022689" type="Const" version="opset1">
+ <data element_type="f16" offset="16" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="19" name="5087509122584" type="Const" version="opset1">
+ <data element_type="f16" offset="18" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="20" name="2704270821135" type="Const" version="opset1">
+ <data element_type="f16" offset="20" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="21" name="2705270921897" type="Const" version="opset1">
+ <data element_type="f16" offset="22" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="22" name="2706271019395" type="Const" version="opset1">
+ <data element_type="f16" offset="20" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="23" name="2707271122200" type="Const" version="opset1">
+ <data element_type="f16" offset="22" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="24" name="3644364819641" type="Const" version="opset1">
+ <data element_type="f16" offset="24" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="25" name="3645364922491" type="Const" version="opset1">
+ <data element_type="f16" offset="26" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="26" name="3646365020877" type="Const" version="opset1">
+ <data element_type="f16" offset="24" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="27" name="3647365121072" type="Const" version="opset1">
+ <data element_type="f16" offset="26" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="28" name="4264426819806" type="Const" version="opset1">
+ <data element_type="f16" offset="28" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="29" name="4265426922113" type="Const" version="opset1">
+ <data element_type="f16" offset="30" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="30" name="4266427022503" type="Const" version="opset1">
+ <data element_type="f16" offset="28" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="31" name="4267427119686" type="Const" version="opset1">
+ <data element_type="f16" offset="30" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="32" name="3544354821201" type="Const" version="opset1">
+ <data element_type="f16" offset="32" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="33" name="3545354922467" type="Const" version="opset1">
+ <data element_type="f16" offset="34" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="34" name="3546355019728" type="Const" version="opset1">
+ <data element_type="f16" offset="32" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="35" name="3547355122629" type="Const" version="opset1">
+ <data element_type="f16" offset="34" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="36" name="4184418820538" type="Const" version="opset1">
+ <data element_type="f16" offset="36" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="37" name="4185418922371" type="Const" version="opset1">
+ <data element_type="f16" offset="38" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="38" name="4186419021999" type="Const" version="opset1">
+ <data element_type="f16" offset="36" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="39" name="4187419121210" type="Const" version="opset1">
+ <data element_type="f16" offset="38" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="40" name="2684268821042" type="Const" version="opset1">
+ <data element_type="f16" offset="40" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="41" name="2685268921624" type="Const" version="opset1">
+ <data element_type="f16" offset="42" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="42" name="2686269019476" type="Const" version="opset1">
+ <data element_type="f16" offset="40" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="43" name="2687269119869" type="Const" version="opset1">
+ <data element_type="f16" offset="42" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="44" name="3964396821630" type="Const" version="opset1">
+ <data element_type="f16" offset="44" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="45" name="3965396921000" type="Const" version="opset1">
+ <data element_type="f16" offset="46" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="46" name="3966397020868" type="Const" version="opset1">
+ <data element_type="f16" offset="44" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="47" name="3967397119716" type="Const" version="opset1">
+ <data element_type="f16" offset="46" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="48" name="3104310821618" type="Const" version="opset1">
+ <data element_type="f16" offset="48" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="49" name="3105310921432" type="Const" version="opset1">
+ <data element_type="f16" offset="50" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="50" name="3106311020412" type="Const" version="opset1">
+ <data element_type="f16" offset="48" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="51" name="3107311119836" type="Const" version="opset1">
+ <data element_type="f16" offset="50" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="52" name="3424342820193" type="Const" version="opset1">
+ <data element_type="f16" offset="52" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="53" name="3425342920448" type="Const" version="opset1">
+ <data element_type="f16" offset="54" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="54" name="3426343022800" type="Const" version="opset1">
+ <data element_type="f16" offset="52" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="55" name="3427343119422" type="Const" version="opset1">
+ <data element_type="f16" offset="54" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="56" name="4464446821750" type="Const" version="opset1">
+ <data element_type="f16" offset="56" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="57" name="4465446919674" type="Const" version="opset1">
+ <data element_type="f16" offset="58" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="58" name="4466447022569" type="Const" version="opset1">
+ <data element_type="f16" offset="56" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="59" name="4467447120724" type="Const" version="opset1">
+ <data element_type="f16" offset="58" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="60" name="4504450820118" type="Const" version="opset1">
+ <data element_type="f16" offset="60" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="61" name="4505450922344" type="Const" version="opset1">
+ <data element_type="f16" offset="62" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="62" name="4506451022254" type="Const" version="opset1">
+ <data element_type="f16" offset="60" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="63" name="4507451120142" type="Const" version="opset1">
+ <data element_type="f16" offset="62" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="64" name="4784478821858" type="Const" version="opset1">
+ <data element_type="f16" offset="64" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="65" name="4785478921141" type="Const" version="opset1">
+ <data element_type="f16" offset="66" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="66" name="4786479020904" type="Const" version="opset1">
+ <data element_type="f16" offset="64" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="67" name="4787479122665" type="Const" version="opset1">
+ <data element_type="f16" offset="66" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="68" name="4664466822413" type="Const" version="opset1">
+ <data element_type="f16" offset="68" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="69" name="4665466922935" type="Const" version="opset1">
+ <data element_type="f16" offset="70" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="70" name="4666467021552" type="Const" version="opset1">
+ <data element_type="f16" offset="68" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="71" name="4667467121786" type="Const" version="opset1">
+ <data element_type="f16" offset="70" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="72" name="3824382821882" type="Const" version="opset1">
+ <data element_type="f16" offset="72" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="73" name="3825382921681" type="Const" version="opset1">
+ <data element_type="f16" offset="74" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="74" name="3826383022695" type="Const" version="opset1">
+ <data element_type="f16" offset="72" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="75" name="3827383122338" type="Const" version="opset1">
+ <data element_type="f16" offset="74" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="76" name="4024402821510" type="Const" version="opset1">
+ <data element_type="f16" offset="76" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="77" name="4025402920178" type="Const" version="opset1">
+ <data element_type="f16" offset="78" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="78" name="4026403019695" type="Const" version="opset1">
+ <data element_type="f16" offset="76" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="79" name="4027403121276" type="Const" version="opset1">
+ <data element_type="f16" offset="78" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="80" name="2624262820919" type="Const" version="opset1">
+ <data element_type="f16" offset="80" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="81" name="2625262919911" type="Const" version="opset1">
+ <data element_type="f16" offset="82" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="82" name="2626263020259" type="Const" version="opset1">
+ <data element_type="f16" offset="80" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="83" name="2627263122752" type="Const" version="opset1">
+ <data element_type="f16" offset="82" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="84" name="3404340820913" type="Const" version="opset1">
+ <data element_type="f16" offset="84" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="85" name="3405340921429" type="Const" version="opset1">
+ <data element_type="f16" offset="86" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="86" name="3406341021609" type="Const" version="opset1">
+ <data element_type="f16" offset="84" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="87" name="3407341122446" type="Const" version="opset1">
+ <data element_type="f16" offset="86" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="88" name="4404440822530" type="Const" version="opset1">
+ <data element_type="f16" offset="88" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="89" name="4405440921777" type="Const" version="opset1">
+ <data element_type="f16" offset="90" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="90" name="4406441021564" type="Const" version="opset1">
+ <data element_type="f16" offset="88" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="91" name="4407441121870" type="Const" version="opset1">
+ <data element_type="f16" offset="90" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="92" name="5204520822269" type="Const" version="opset1">
+ <data element_type="f16" offset="92" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="93" name="5205520922212" type="Const" version="opset1">
+ <data element_type="f16" offset="94" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="94" name="5206521022803" type="Const" version="opset1">
+ <data element_type="f16" offset="92" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="95" name="5207521121507" type="Const" version="opset1">
+ <data element_type="f16" offset="94" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="96" name="4944494822152" type="Const" version="opset1">
+ <data element_type="f16" offset="96" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="97" name="4945494919998" type="Const" version="opset1">
+ <data element_type="f16" offset="98" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="98" name="4946495021855" type="Const" version="opset1">
+ <data element_type="f16" offset="96" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="99" name="4947495121708" type="Const" version="opset1">
+ <data element_type="f16" offset="98" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="100" name="4484448821936" type="Const" version="opset1">
+ <data element_type="f16" offset="100" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="101" name="4485448922644" type="Const" version="opset1">
+ <data element_type="f16" offset="102" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="102" name="4486449019917" type="Const" version="opset1">
+ <data element_type="f16" offset="100" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="103" name="4487449119587" type="Const" version="opset1">
+ <data element_type="f16" offset="102" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="104" name="2744274822863" type="Const" version="opset1">
+ <data element_type="f16" offset="104" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="105" name="2745274921102" type="Const" version="opset1">
+ <data element_type="f16" offset="106" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="106" name="2746275020253" type="Const" version="opset1">
+ <data element_type="f16" offset="104" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="107" name="2747275121492" type="Const" version="opset1">
+ <data element_type="f16" offset="106" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="108" name="4444444819962" type="Const" version="opset1">
+ <data element_type="f16" offset="108" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="109" name="4445444920199" type="Const" version="opset1">
+ <data element_type="f16" offset="110" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="110" name="4446445022824" type="Const" version="opset1">
+ <data element_type="f16" offset="108" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="111" name="4447445119440" type="Const" version="opset1">
+ <data element_type="f16" offset="110" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="112" name="2904290821645" type="Const" version="opset1">
+ <data element_type="f16" offset="112" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="113" name="2905290920598" type="Const" version="opset1">
+ <data element_type="f16" offset="114" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="114" name="2906291022980" type="Const" version="opset1">
+ <data element_type="f16" offset="112" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="115" name="2907291122245" type="Const" version="opset1">
+ <data element_type="f16" offset="114" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="116" name="5044504820001" type="Const" version="opset1">
+ <data element_type="f16" offset="116" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="117" name="5045504922119" type="Const" version="opset1">
+ <data element_type="f16" offset="118" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="118" name="5046505019875" type="Const" version="opset1">
+ <data element_type="f16" offset="116" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="119" name="5047505120280" type="Const" version="opset1">
+ <data element_type="f16" offset="118" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="120" name="5024502822506" type="Const" version="opset1">
+ <data element_type="f16" offset="120" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="121" name="5025502922287" type="Const" version="opset1">
+ <data element_type="f16" offset="122" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="122" name="5026503020586" type="Const" version="opset1">
+ <data element_type="f16" offset="120" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="123" name="5027503119461" type="Const" version="opset1">
+ <data element_type="f16" offset="122" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="124" name="5404540821597" type="Const" version="opset1">
+ <data element_type="f16" offset="124" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="125" name="5405540920799" type="Const" version="opset1">
+ <data element_type="f16" offset="126" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="126" name="5406541022485" type="Const" version="opset1">
+ <data element_type="f16" offset="124" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="127" name="5407541122536" type="Const" version="opset1">
+ <data element_type="f16" offset="126" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="128" name="4964496821996" type="Const" version="opset1">
+ <data element_type="f16" offset="128" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="129" name="4965496922401" type="Const" version="opset1">
+ <data element_type="f16" offset="130" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="130" name="4966497020568" type="Const" version="opset1">
+ <data element_type="f16" offset="128" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="131" name="4967497122734" type="Const" version="opset1">
+ <data element_type="f16" offset="130" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="132" name="5304530820634" type="Const" version="opset1">
+ <data element_type="f16" offset="132" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="133" name="5305530919428" type="Const" version="opset1">
+ <data element_type="f16" offset="134" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="134" name="5306531021435" type="Const" version="opset1">
+ <data element_type="f16" offset="132" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="135" name="5307531121369" type="Const" version="opset1">
+ <data element_type="f16" offset="134" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="136" name="2724272819872" type="Const" version="opset1">
+ <data element_type="f16" offset="136" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="137" name="2725272921912" type="Const" version="opset1">
+ <data element_type="f16" offset="138" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="138" name="2726273022416" type="Const" version="opset1">
+ <data element_type="f16" offset="136" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="139" name="2727273122437" type="Const" version="opset1">
+ <data element_type="f16" offset="138" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="140" name="3324332819809" type="Const" version="opset1">
+ <data element_type="f16" offset="140" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="141" name="3325332922389" type="Const" version="opset1">
+ <data element_type="f16" offset="142" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="142" name="3326333020481" type="Const" version="opset1">
+ <data element_type="f16" offset="140" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="143" name="3327333120748" type="Const" version="opset1">
+ <data element_type="f16" offset="142" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="144" name="4004400822971" type="Const" version="opset1">
+ <data element_type="f16" offset="144" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="145" name="4005400921327" type="Const" version="opset1">
+ <data element_type="f16" offset="146" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="146" name="4006401020685" type="Const" version="opset1">
+ <data element_type="f16" offset="144" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="147" name="4007401120556" type="Const" version="opset1">
+ <data element_type="f16" offset="146" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="148" name="4764476820811" type="Const" version="opset1">
+ <data element_type="f16" offset="148" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="149" name="4765476921228" type="Const" version="opset1">
+ <data element_type="f16" offset="150" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="150" name="4766477020508" type="Const" version="opset1">
+ <data element_type="f16" offset="148" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="151" name="4767477120679" type="Const" version="opset1">
+ <data element_type="f16" offset="150" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="152" name="2824282819434" type="Const" version="opset1">
+ <data element_type="f16" offset="152" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="153" name="2825282922851" type="Const" version="opset1">
+ <data element_type="f16" offset="154" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="154" name="2826283021987" type="Const" version="opset1">
+ <data element_type="f16" offset="152" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="155" name="2827283122194" type="Const" version="opset1">
+ <data element_type="f16" offset="154" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="156" name="4064406822764" type="Const" version="opset1">
+ <data element_type="f16" offset="156" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="157" name="4065406921414" type="Const" version="opset1">
+ <data element_type="f16" offset="158" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="158" name="4066407019647" type="Const" version="opset1">
+ <data element_type="f16" offset="156" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="159" name="4067407121186" type="Const" version="opset1">
+ <data element_type="f16" offset="158" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="160" name="2984298821153" type="Const" version="opset1">
+ <data element_type="f16" offset="160" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="161" name="2985298920082" type="Const" version="opset1">
+ <data element_type="f16" offset="162" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="162" name="2986299020013" type="Const" version="opset1">
+ <data element_type="f16" offset="160" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="163" name="2987299122623" type="Const" version="opset1">
+ <data element_type="f16" offset="162" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="164" name="data" type="Parameter" version="opset1">
+ <data element_type="f16" shape="1,3,320,544"/>
+ <output>
+ <port id="0" names="data" precision="FP16">
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>320</dim>
+ <dim>544</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="165" name="data_mul_2364474720739" type="Const" version="opset1">
+ <data element_type="f16" offset="164" shape="1,3,1,1" size="6"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="166" name="data/norm/bn/mean/Fused_Mul_" type="Multiply" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>320</dim>
+ <dim>544</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>320</dim>
+ <dim>544</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="167" name="data_add_2364674920346" type="Const" version="opset1">
+ <data element_type="f16" offset="170" shape="1,3,1,1" size="6"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="168" name="data/norm/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>320</dim>
+ <dim>544</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="data/norm/bn" precision="FP16">
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>320</dim>
+ <dim>544</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="169" name="init_block1/dim_inc/conv/fq_input_0" type="FakeQuantize" version="opset1">
+ <data auto_broadcast="numpy" levels="256"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>320</dim>
+ <dim>544</dim>
+ </port>
+ <port id="1"/>
+ <port id="2"/>
+ <port id="3"/>
+ <port id="4"/>
+ </input>
+ <output>
+ <port id="5" precision="FP16">
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>320</dim>
+ <dim>544</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="170" name="init_block1/dim_inc/bn/mean/Fused_Mul__copy75110080/quantized1156020379" type="Const" version="opset1">
+ <data element_type="i8" offset="176" shape="32,3,3,3" size="864"/>
+ <output>
+ <port id="0" precision="I8">
+ <dim>32</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="171" name="init_block1/dim_inc/bn/mean/Fused_Mul__copy75110080/quantized/to_f16" type="Convert" version="opset1">
+ <data destination_type="f16"/>
+ <input>
+ <port id="0">
+ <dim>32</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </input>
+ <output>
+ <port id="1" precision="FP16">
+ <dim>32</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="172" name="init_block1/dim_inc/conv/fq_weights_1/zero_point1157320874" type="Const" version="opset1">
+ <data element_type="f16" offset="1040" shape="32,1,1,1" size="64"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="173" name="init_block1/dim_inc/conv/fq_weights_1/minus_zp" type="Subtract" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>32</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ <port id="1">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>32</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="174" name="init_block1/dim_inc/conv/fq_weights_1/scale1156821582" type="Const" version="opset1">
+ <data element_type="f16" offset="1104" shape="32,1,1,1" size="64"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="175" name="init_block1/dim_inc/conv/fq_weights_1/mulpiply_by_scale" type="Multiply" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>32</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ <port id="1">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>32</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="176" name="init_block1/dim_inc/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1,1" pads_begin="1,1" pads_end="1,1" strides="2,2"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>320</dim>
+ <dim>544</dim>
+ </port>
+ <port id="1">
+ <dim>32</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="177" name="data_add_236492365475319752" type="Const" version="opset1">
+ <data element_type="f16" offset="1168" shape="1,32,1,1" size="64"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="178" name="init_block1/dim_inc/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="init_block1/dim_inc/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="179" name="init_block1/dim_inc/fn" type="ReLU" version="opset1">
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ </input>
+ <output>
+ <port id="1" names="init_block1/dim_inc/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="180" name="bottleneck1_1/add/fq_input_0" type="FakeQuantize" version="opset1">
+ <data auto_broadcast="numpy" levels="256"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ <port id="1"/>
+ <port id="2"/>
+ <port id="3"/>
+ <port id="4"/>
+ </input>
+ <output>
+ <port id="5" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="181" name="4074407821579" type="Const" version="opset1">
+ <data element_type="f16" offset="1232" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="182" name="4075407920544" type="Const" version="opset1">
+ <data element_type="f16" offset="1234" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="183" name="4076408020457" type="Const" version="opset1">
+ <data element_type="f16" offset="1232" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="184" name="4077408119959" type="Const" version="opset1">
+ <data element_type="f16" offset="1234" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="185" name="5484548819965" type="Const" version="opset1">
+ <data element_type="f16" offset="1236" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="186" name="5485548920790" type="Const" version="opset1">
+ <data element_type="f16" offset="1238" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="187" name="5486549020364" type="Const" version="opset1">
+ <data element_type="f16" offset="1236" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="188" name="5487549119788" type="Const" version="opset1">
+ <data element_type="f16" offset="1238" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="189" name="3844384820577" type="Const" version="opset1">
+ <data element_type="f16" offset="1240" shape="1,8,1,1" size="16"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="190" name="3845384920994" type="Const" version="opset1">
+ <data element_type="f16" offset="1256" shape="1,8,1,1" size="16"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="191" name="3846385019611" type="Const" version="opset1">
+ <data element_type="f16" offset="1240" shape="1,8,1,1" size="16"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="192" name="3847385119518" type="Const" version="opset1">
+ <data element_type="f16" offset="1256" shape="1,8,1,1" size="16"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="193" name="bottleneck1_1/dim_red/bn/mean/Fused_Mul__copy75610082/quantized1376820805" type="Const" version="opset1">
+ <data element_type="i8" offset="1272" shape="8,32,1,1" size="256"/>
+ <output>
+ <port id="0" precision="I8">
+ <dim>8</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="194" name="bottleneck1_1/dim_red/bn/mean/Fused_Mul__copy75610082/quantized/to_f16" type="Convert" version="opset1">
+ <data destination_type="f16"/>
+ <input>
+ <port id="0">
+ <dim>8</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="1" precision="FP16">
+ <dim>8</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="195" name="bottleneck1_1/dim_red/conv/fq_weights_1/zero_point1378120046" type="Const" version="opset1">
+ <data element_type="f16" offset="1528" shape="8,1,1,1" size="16"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="196" name="bottleneck1_1/dim_red/conv/fq_weights_1/minus_zp" type="Subtract" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>8</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="1">
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>8</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="197" name="bottleneck1_1/dim_red/conv/fq_weights_1/scale1377620769" type="Const" version="opset1">
+ <data element_type="f16" offset="1544" shape="8,1,1,1" size="16"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="198" name="bottleneck1_1/dim_red/conv/fq_weights_1/mulpiply_by_scale" type="Multiply" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>8</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="1">
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>8</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="199" name="bottleneck1_1/dim_red/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1,1" pads_begin="0,0" pads_end="0,0" strides="1,1"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ <port id="1">
+ <dim>8</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="200" name="data_add_236572366275820892" type="Const" version="opset1">
+ <data element_type="f16" offset="1560" shape="1,8,1,1" size="16"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="201" name="bottleneck1_1/dim_red/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck1_1/dim_red/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="202" name="bottleneck1_1/dim_red/fn/weights3078040028760" type="Const" version="opset1">
+ <data element_type="f32" offset="1576" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="203" name="bottleneck1_1/dim_red/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck1_1/dim_red/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="204" name="bottleneck1_1/inner/dw1/conv/fq_input_0" type="FakeQuantize" version="opset1">
+ <data auto_broadcast="numpy" levels="256"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="2">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="3">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="4">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="5" precision="FP16">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="205" name="16855/value1685720643" type="Const" version="opset1">
+ <data element_type="i64" offset="1580" shape="5" size="40"/>
+ <output>
+ <port id="0" precision="I64">
+ <dim>5</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="206" name="bottleneck1_1/inner/dw1/bn/mean/Fused_Mul__copy76210085/quantized1374422218" type="Const" version="opset1">
+ <data element_type="i8" offset="1620" shape="8,1,3,3" size="72"/>
+ <output>
+ <port id="0" precision="I8">
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="207" name="bottleneck1_1/inner/dw1/bn/mean/Fused_Mul__copy76210085/quantized/to_f16" type="Convert" version="opset1">
+ <data destination_type="f16"/>
+ <input>
+ <port id="0">
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </input>
+ <output>
+ <port id="1" precision="FP16">
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="208" name="bottleneck1_1/inner/dw1/conv/fq_weights_1/zero_point1375719575" type="Const" version="opset1">
+ <data element_type="f16" offset="1692" shape="8,1,1,1" size="16"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="209" name="bottleneck1_1/inner/dw1/conv/fq_weights_1/minus_zp" type="Subtract" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ <port id="1">
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="210" name="bottleneck1_1/inner/dw1/conv/fq_weights_1/scale1375220715" type="Const" version="opset1">
+ <data element_type="f16" offset="1708" shape="8,1,1,1" size="16"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="211" name="bottleneck1_1/inner/dw1/conv/fq_weights_1/mulpiply_by_scale" type="Multiply" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ <port id="1">
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="212" name="16855" type="Reshape" version="opset1">
+ <data special_zero="true"/>
+ <input>
+ <port id="0">
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ <port id="1">
+ <dim>5</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="213" name="bottleneck1_1/inner/dw1/conv" type="GroupConvolution" version="opset1">
+ <data auto_pad="explicit" dilations="1,1" pads_begin="1,1" pads_end="1,1" strides="1,1"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ <port id="1">
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="214" name="data_add_236652367076419749" type="Const" version="opset1">
+ <data element_type="f16" offset="1724" shape="1,8,1,1" size="16"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="215" name="bottleneck1_1/inner/dw1/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck1_1/inner/dw1/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="216" name="bottleneck1_1/inner/dw1/fn/weights3100440070766" type="Const" version="opset1">
+ <data element_type="f32" offset="1576" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="217" name="bottleneck1_1/inner/dw1/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck1_1/inner/dw1/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="218" name="bottleneck1_1/dim_inc/conv/fq_input_0" type="FakeQuantize" version="opset1">
+ <data auto_broadcast="numpy" levels="256"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ <port id="1"/>
+ <port id="2"/>
+ <port id="3"/>
+ <port id="4"/>
+ </input>
+ <output>
+ <port id="5" precision="FP16">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="219" name="bottleneck1_1/dim_inc/bn/mean/Fused_Mul__copy76810088/quantized1280822266" type="Const" version="opset1">
+ <data element_type="i8" offset="1740" shape="32,8,1,1" size="256"/>
+ <output>
+ <port id="0" precision="I8">
+ <dim>32</dim>
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="220" name="bottleneck1_1/dim_inc/bn/mean/Fused_Mul__copy76810088/quantized/to_f16" type="Convert" version="opset1">
+ <data destination_type="f16"/>
+ <input>
+ <port id="0">
+ <dim>32</dim>
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="1" precision="FP16">
+ <dim>32</dim>
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="221" name="bottleneck1_1/dim_inc/conv/fq_weights_1/zero_point1282122905" type="Const" version="opset1">
+ <data element_type="f16" offset="1996" shape="32,1,1,1" size="64"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="222" name="bottleneck1_1/dim_inc/conv/fq_weights_1/minus_zp" type="Subtract" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>32</dim>
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="1">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>32</dim>
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="223" name="bottleneck1_1/dim_inc/conv/fq_weights_1/scale1281622482" type="Const" version="opset1">
+ <data element_type="f16" offset="2060" shape="32,1,1,1" size="64"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="224" name="bottleneck1_1/dim_inc/conv/fq_weights_1/mulpiply_by_scale" type="Multiply" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>32</dim>
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="1">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>32</dim>
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="225" name="bottleneck1_1/dim_inc/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1,1" pads_begin="0,0" pads_end="0,0" strides="1,1"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ <port id="1">
+ <dim>32</dim>
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="226" name="data_add_236732367877021738" type="Const" version="opset1">
+ <data element_type="f16" offset="2124" shape="1,32,1,1" size="64"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="227" name="bottleneck1_1/dim_inc/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck1_1/dim_inc/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="228" name="bottleneck1_1/add/fq_input_1" type="FakeQuantize" version="opset1">
+ <data auto_broadcast="numpy" levels="256"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ <port id="1"/>
+ <port id="2"/>
+ <port id="3"/>
+ <port id="4"/>
+ </input>
+ <output>
+ <port id="5" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="229" name="bottleneck1_1/add" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck1_1/add" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="230" name="bottleneck1_1/fn/weights3085240283773" type="Const" version="opset1">
+ <data element_type="f32" offset="1576" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="231" name="bottleneck1_1/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck1_1/add" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="232" name="bottleneck1_2/add/fq_input_0" type="FakeQuantize" version="opset1">
+ <data auto_broadcast="numpy" levels="256"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ <port id="1"/>
+ <port id="2"/>
+ <port id="3"/>
+ <port id="4"/>
+ </input>
+ <output>
+ <port id="5" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="233" name="2834283820706" type="Const" version="opset1">
+ <data element_type="f16" offset="2188" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="234" name="2835283922026" type="Const" version="opset1">
+ <data element_type="f16" offset="2190" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="235" name="2836284020475" type="Const" version="opset1">
+ <data element_type="f16" offset="2188" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="236" name="2837284120844" type="Const" version="opset1">
+ <data element_type="f16" offset="2190" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="237" name="3604360820832" type="Const" version="opset1">
+ <data element_type="f16" offset="2192" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="238" name="3605360921774" type="Const" version="opset1">
+ <data element_type="f16" offset="2194" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="239" name="3606361021057" type="Const" version="opset1">
+ <data element_type="f16" offset="2192" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="240" name="3607361121939" type="Const" version="opset1">
+ <data element_type="f16" offset="2194" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="241" name="4324432819458" type="Const" version="opset1">
+ <data element_type="f16" offset="2196" shape="1,8,1,1" size="16"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="242" name="4325432919383" type="Const" version="opset1">
+ <data element_type="f16" offset="2212" shape="1,8,1,1" size="16"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="243" name="4326433019566" type="Const" version="opset1">
+ <data element_type="f16" offset="2196" shape="1,8,1,1" size="16"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="244" name="4327433121792" type="Const" version="opset1">
+ <data element_type="f16" offset="2212" shape="1,8,1,1" size="16"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="245" name="bottleneck1_2/dim_red/bn/mean/Fused_Mul__copy77510091/quantized1273622236" type="Const" version="opset1">
+ <data element_type="i8" offset="2228" shape="8,32,1,1" size="256"/>
+ <output>
+ <port id="0" precision="I8">
+ <dim>8</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="246" name="bottleneck1_2/dim_red/bn/mean/Fused_Mul__copy77510091/quantized/to_f16" type="Convert" version="opset1">
+ <data destination_type="f16"/>
+ <input>
+ <port id="0">
+ <dim>8</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="1" precision="FP16">
+ <dim>8</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="247" name="bottleneck1_2/dim_red/conv/fq_weights_1/zero_point1274922707" type="Const" version="opset1">
+ <data element_type="f16" offset="2484" shape="8,1,1,1" size="16"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="248" name="bottleneck1_2/dim_red/conv/fq_weights_1/minus_zp" type="Subtract" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>8</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="1">
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>8</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="249" name="bottleneck1_2/dim_red/conv/fq_weights_1/scale1274420130" type="Const" version="opset1">
+ <data element_type="f16" offset="2500" shape="8,1,1,1" size="16"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="250" name="bottleneck1_2/dim_red/conv/fq_weights_1/mulpiply_by_scale" type="Multiply" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>8</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="1">
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>8</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="251" name="bottleneck1_2/dim_red/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1,1" pads_begin="0,0" pads_end="0,0" strides="1,1"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ <port id="1">
+ <dim>8</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="252" name="data_add_236812368677719602" type="Const" version="opset1">
+ <data element_type="f16" offset="2516" shape="1,8,1,1" size="16"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="253" name="bottleneck1_2/dim_red/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck1_2/dim_red/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="254" name="bottleneck1_2/dim_red/fn/weights3083240265779" type="Const" version="opset1">
+ <data element_type="f32" offset="1576" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="255" name="bottleneck1_2/dim_red/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck1_2/dim_red/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="256" name="bottleneck1_2/inner/dw1/conv/fq_input_0" type="FakeQuantize" version="opset1">
+ <data auto_broadcast="numpy" levels="256"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="2">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="3">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="4">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="5" precision="FP16">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="257" name="16887/value1688920700" type="Const" version="opset1">
+ <data element_type="i64" offset="1580" shape="5" size="40"/>
+ <output>
+ <port id="0" precision="I64">
+ <dim>5</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="258" name="bottleneck1_2/inner/dw1/bn/mean/Fused_Mul__copy78110094/quantized1196822821" type="Const" version="opset1">
+ <data element_type="i8" offset="2532" shape="8,1,3,3" size="72"/>
+ <output>
+ <port id="0" precision="I8">
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="259" name="bottleneck1_2/inner/dw1/bn/mean/Fused_Mul__copy78110094/quantized/to_f16" type="Convert" version="opset1">
+ <data destination_type="f16"/>
+ <input>
+ <port id="0">
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </input>
+ <output>
+ <port id="1" precision="FP16">
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="260" name="bottleneck1_2/inner/dw1/conv/fq_weights_1/zero_point1198120871" type="Const" version="opset1">
+ <data element_type="f16" offset="2604" shape="8,1,1,1" size="16"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="261" name="bottleneck1_2/inner/dw1/conv/fq_weights_1/minus_zp" type="Subtract" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ <port id="1">
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="262" name="bottleneck1_2/inner/dw1/conv/fq_weights_1/scale1197621684" type="Const" version="opset1">
+ <data element_type="f16" offset="2620" shape="8,1,1,1" size="16"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="263" name="bottleneck1_2/inner/dw1/conv/fq_weights_1/mulpiply_by_scale" type="Multiply" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ <port id="1">
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="264" name="16887" type="Reshape" version="opset1">
+ <data special_zero="true"/>
+ <input>
+ <port id="0">
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ <port id="1">
+ <dim>5</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="265" name="bottleneck1_2/inner/dw1/conv" type="GroupConvolution" version="opset1">
+ <data auto_pad="explicit" dilations="1,1" pads_begin="1,1" pads_end="1,1" strides="1,1"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ <port id="1">
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="266" name="data_add_236892369478320910" type="Const" version="opset1">
+ <data element_type="f16" offset="2636" shape="1,8,1,1" size="16"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="267" name="bottleneck1_2/inner/dw1/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck1_2/inner/dw1/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="268" name="bottleneck1_2/inner/dw1/fn/weights3082440295785" type="Const" version="opset1">
+ <data element_type="f32" offset="1576" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="269" name="bottleneck1_2/inner/dw1/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck1_2/inner/dw1/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="270" name="bottleneck1_2/dim_inc/conv/fq_input_0" type="FakeQuantize" version="opset1">
+ <data auto_broadcast="numpy" levels="256"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ <port id="1"/>
+ <port id="2"/>
+ <port id="3"/>
+ <port id="4"/>
+ </input>
+ <output>
+ <port id="5" precision="FP16">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="271" name="bottleneck1_2/dim_inc/bn/mean/Fused_Mul__copy78710097/quantized1285621807" type="Const" version="opset1">
+ <data element_type="i8" offset="2652" shape="32,8,1,1" size="256"/>
+ <output>
+ <port id="0" precision="I8">
+ <dim>32</dim>
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="272" name="bottleneck1_2/dim_inc/bn/mean/Fused_Mul__copy78710097/quantized/to_f16" type="Convert" version="opset1">
+ <data destination_type="f16"/>
+ <input>
+ <port id="0">
+ <dim>32</dim>
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="1" precision="FP16">
+ <dim>32</dim>
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="273" name="bottleneck1_2/dim_inc/conv/fq_weights_1/zero_point1286920469" type="Const" version="opset1">
+ <data element_type="f16" offset="2908" shape="32,1,1,1" size="64"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="274" name="bottleneck1_2/dim_inc/conv/fq_weights_1/minus_zp" type="Subtract" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>32</dim>
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="1">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>32</dim>
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="275" name="bottleneck1_2/dim_inc/conv/fq_weights_1/scale1286419986" type="Const" version="opset1">
+ <data element_type="f16" offset="2972" shape="32,1,1,1" size="64"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="276" name="bottleneck1_2/dim_inc/conv/fq_weights_1/mulpiply_by_scale" type="Multiply" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>32</dim>
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="1">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>32</dim>
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="277" name="bottleneck1_2/dim_inc/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1,1" pads_begin="0,0" pads_end="0,0" strides="1,1"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ <port id="1">
+ <dim>32</dim>
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="278" name="data_add_236972370278919596" type="Const" version="opset1">
+ <data element_type="f16" offset="3036" shape="1,32,1,1" size="64"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="279" name="bottleneck1_2/dim_inc/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck1_2/dim_inc/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="280" name="bottleneck1_2/add/fq_input_1" type="FakeQuantize" version="opset1">
+ <data auto_broadcast="numpy" levels="256"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ <port id="1"/>
+ <port id="2"/>
+ <port id="3"/>
+ <port id="4"/>
+ </input>
+ <output>
+ <port id="5" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="281" name="bottleneck1_2/add" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck1_2/add" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="282" name="bottleneck1_2/fn/weights3106840700792" type="Const" version="opset1">
+ <data element_type="f32" offset="1576" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="283" name="bottleneck1_2/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck1_2/add" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="284" name="bottleneck1_3/add/fq_input_0" type="FakeQuantize" version="opset1">
+ <data auto_broadcast="numpy" levels="256"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ <port id="1"/>
+ <port id="2"/>
+ <port id="3"/>
+ <port id="4"/>
+ </input>
+ <output>
+ <port id="5" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="285" name="4774477819857" type="Const" version="opset1">
+ <data element_type="f16" offset="3100" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="286" name="4775477922080" type="Const" version="opset1">
+ <data element_type="f16" offset="3102" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="287" name="4776478019659" type="Const" version="opset1">
+ <data element_type="f16" offset="3100" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="288" name="4777478120052" type="Const" version="opset1">
+ <data element_type="f16" offset="3102" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="289" name="2784278821840" type="Const" version="opset1">
+ <data element_type="f16" offset="3104" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="290" name="2785278921474" type="Const" version="opset1">
+ <data element_type="f16" offset="3106" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="291" name="2786279021399" type="Const" version="opset1">
+ <data element_type="f16" offset="3104" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="292" name="2787279122146" type="Const" version="opset1">
+ <data element_type="f16" offset="3106" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="293" name="2864286822065" type="Const" version="opset1">
+ <data element_type="f16" offset="3108" shape="1,8,1,1" size="16"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="294" name="2865286922521" type="Const" version="opset1">
+ <data element_type="f16" offset="3124" shape="1,8,1,1" size="16"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="295" name="2866287022917" type="Const" version="opset1">
+ <data element_type="f16" offset="3108" shape="1,8,1,1" size="16"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="296" name="2867287122137" type="Const" version="opset1">
+ <data element_type="f16" offset="3124" shape="1,8,1,1" size="16"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="297" name="bottleneck1_3/dim_red/bn/mean/Fused_Mul__copy79410100/quantized1420021693" type="Const" version="opset1">
+ <data element_type="i8" offset="3140" shape="8,32,1,1" size="256"/>
+ <output>
+ <port id="0" precision="I8">
+ <dim>8</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="298" name="bottleneck1_3/dim_red/bn/mean/Fused_Mul__copy79410100/quantized/to_f16" type="Convert" version="opset1">
+ <data destination_type="f16"/>
+ <input>
+ <port id="0">
+ <dim>8</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="1" precision="FP16">
+ <dim>8</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="299" name="bottleneck1_3/dim_red/conv/fq_weights_1/zero_point1421320478" type="Const" version="opset1">
+ <data element_type="f16" offset="3396" shape="8,1,1,1" size="16"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="300" name="bottleneck1_3/dim_red/conv/fq_weights_1/minus_zp" type="Subtract" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>8</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="1">
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>8</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="301" name="bottleneck1_3/dim_red/conv/fq_weights_1/scale1420819449" type="Const" version="opset1">
+ <data element_type="f16" offset="3412" shape="8,1,1,1" size="16"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="302" name="bottleneck1_3/dim_red/conv/fq_weights_1/mulpiply_by_scale" type="Multiply" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>8</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="1">
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>8</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="303" name="bottleneck1_3/dim_red/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1,1" pads_begin="0,0" pads_end="0,0" strides="1,1"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ <port id="1">
+ <dim>8</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="304" name="data_add_237052371079621483" type="Const" version="opset1">
+ <data element_type="f16" offset="3428" shape="1,8,1,1" size="16"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="305" name="bottleneck1_3/dim_red/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck1_3/dim_red/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="306" name="bottleneck1_3/dim_red/fn/weights3116439665798" type="Const" version="opset1">
+ <data element_type="f32" offset="1576" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="307" name="bottleneck1_3/dim_red/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck1_3/dim_red/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="308" name="bottleneck1_3/inner/dw1/conv/fq_input_0" type="FakeQuantize" version="opset1">
+ <data auto_broadcast="numpy" levels="256"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="2">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="3">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="4">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="5" precision="FP16">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="309" name="16807/value1680921342" type="Const" version="opset1">
+ <data element_type="i64" offset="1580" shape="5" size="40"/>
+ <output>
+ <port id="0" precision="I64">
+ <dim>5</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="310" name="bottleneck1_3/inner/dw1/bn/mean/Fused_Mul__copy80010103/quantized1386420673" type="Const" version="opset1">
+ <data element_type="i8" offset="3444" shape="8,1,3,3" size="72"/>
+ <output>
+ <port id="0" precision="I8">
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="311" name="bottleneck1_3/inner/dw1/bn/mean/Fused_Mul__copy80010103/quantized/to_f16" type="Convert" version="opset1">
+ <data destination_type="f16"/>
+ <input>
+ <port id="0">
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </input>
+ <output>
+ <port id="1" precision="FP16">
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="312" name="bottleneck1_3/inner/dw1/conv/fq_weights_1/zero_point1387720220" type="Const" version="opset1">
+ <data element_type="f16" offset="3516" shape="8,1,1,1" size="16"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="313" name="bottleneck1_3/inner/dw1/conv/fq_weights_1/minus_zp" type="Subtract" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ <port id="1">
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="314" name="bottleneck1_3/inner/dw1/conv/fq_weights_1/scale1387221195" type="Const" version="opset1">
+ <data element_type="f16" offset="3532" shape="8,1,1,1" size="16"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="315" name="bottleneck1_3/inner/dw1/conv/fq_weights_1/mulpiply_by_scale" type="Multiply" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ <port id="1">
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="316" name="16807" type="Reshape" version="opset1">
+ <data special_zero="true"/>
+ <input>
+ <port id="0">
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ <port id="1">
+ <dim>5</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="317" name="bottleneck1_3/inner/dw1/conv" type="GroupConvolution" version="opset1">
+ <data auto_pad="explicit" dilations="1,1" pads_begin="1,1" pads_end="1,1" strides="1,1"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ <port id="1">
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="318" name="data_add_237132371880222077" type="Const" version="opset1">
+ <data element_type="f16" offset="3548" shape="1,8,1,1" size="16"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="319" name="bottleneck1_3/inner/dw1/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck1_3/inner/dw1/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="320" name="bottleneck1_3/inner/dw1/fn/weights3087239692804" type="Const" version="opset1">
+ <data element_type="f32" offset="1576" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="321" name="bottleneck1_3/inner/dw1/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck1_3/inner/dw1/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="322" name="bottleneck1_3/dim_inc/conv/fq_input_0" type="FakeQuantize" version="opset1">
+ <data auto_broadcast="numpy" levels="256"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ <port id="1"/>
+ <port id="2"/>
+ <port id="3"/>
+ <port id="4"/>
+ </input>
+ <output>
+ <port id="5" precision="FP16">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="323" name="bottleneck1_3/dim_inc/bn/mean/Fused_Mul__copy80610106/quantized1312021546" type="Const" version="opset1">
+ <data element_type="i8" offset="3564" shape="32,8,1,1" size="256"/>
+ <output>
+ <port id="0" precision="I8">
+ <dim>32</dim>
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="324" name="bottleneck1_3/dim_inc/bn/mean/Fused_Mul__copy80610106/quantized/to_f16" type="Convert" version="opset1">
+ <data destination_type="f16"/>
+ <input>
+ <port id="0">
+ <dim>32</dim>
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="1" precision="FP16">
+ <dim>32</dim>
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="325" name="bottleneck1_3/dim_inc/conv/fq_weights_1/zero_point1313321846" type="Const" version="opset1">
+ <data element_type="f16" offset="3820" shape="32,1,1,1" size="64"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="326" name="bottleneck1_3/dim_inc/conv/fq_weights_1/minus_zp" type="Subtract" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>32</dim>
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="1">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>32</dim>
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="327" name="bottleneck1_3/dim_inc/conv/fq_weights_1/scale1312822020" type="Const" version="opset1">
+ <data element_type="f16" offset="3884" shape="32,1,1,1" size="64"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="328" name="bottleneck1_3/dim_inc/conv/fq_weights_1/mulpiply_by_scale" type="Multiply" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>32</dim>
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="1">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>32</dim>
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="329" name="bottleneck1_3/dim_inc/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1,1" pads_begin="0,0" pads_end="0,0" strides="1,1"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ <port id="1">
+ <dim>32</dim>
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="330" name="data_add_237212372680822044" type="Const" version="opset1">
+ <data element_type="f16" offset="3948" shape="1,32,1,1" size="64"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="331" name="bottleneck1_3/dim_inc/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck1_3/dim_inc/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="332" name="bottleneck1_3/add/fq_input_1" type="FakeQuantize" version="opset1">
+ <data auto_broadcast="numpy" levels="256"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ <port id="1"/>
+ <port id="2"/>
+ <port id="3"/>
+ <port id="4"/>
+ </input>
+ <output>
+ <port id="5" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="333" name="bottleneck1_3/add" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck1_3/add" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="334" name="bottleneck1_3/fn/weights3115639653811" type="Const" version="opset1">
+ <data element_type="f32" offset="1576" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="335" name="bottleneck1_3/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck1_3/add" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="336" name="bottleneck1_4/add/fq_input_0" type="FakeQuantize" version="opset1">
+ <data auto_broadcast="numpy" levels="256"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ <port id="1"/>
+ <port id="2"/>
+ <port id="3"/>
+ <port id="4"/>
+ </input>
+ <output>
+ <port id="5" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="337" name="4014401819608" type="Const" version="opset1">
+ <data element_type="f16" offset="4012" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="338" name="4015401921417" type="Const" version="opset1">
+ <data element_type="f16" offset="4014" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="339" name="4016402019419" type="Const" version="opset1">
+ <data element_type="f16" offset="4012" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="340" name="4017402122053" type="Const" version="opset1">
+ <data element_type="f16" offset="4014" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="341" name="5004500820823" type="Const" version="opset1">
+ <data element_type="f16" offset="4016" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="342" name="5005500920274" type="Const" version="opset1">
+ <data element_type="f16" offset="4018" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="343" name="5006501021309" type="Const" version="opset1">
+ <data element_type="f16" offset="4016" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="344" name="5007501121627" type="Const" version="opset1">
+ <data element_type="f16" offset="4018" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="345" name="3124312820454" type="Const" version="opset1">
+ <data element_type="f16" offset="4020" shape="1,8,1,1" size="16"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="346" name="3125312921702" type="Const" version="opset1">
+ <data element_type="f16" offset="4036" shape="1,8,1,1" size="16"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="347" name="3126313020085" type="Const" version="opset1">
+ <data element_type="f16" offset="4020" shape="1,8,1,1" size="16"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="348" name="3127313119653" type="Const" version="opset1">
+ <data element_type="f16" offset="4036" shape="1,8,1,1" size="16"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="349" name="bottleneck1_4/dim_red/bn/mean/Fused_Mul__copy81310109/quantized1278420061" type="Const" version="opset1">
+ <data element_type="i8" offset="4052" shape="8,32,1,1" size="256"/>
+ <output>
+ <port id="0" precision="I8">
+ <dim>8</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="350" name="bottleneck1_4/dim_red/bn/mean/Fused_Mul__copy81310109/quantized/to_f16" type="Convert" version="opset1">
+ <data destination_type="f16"/>
+ <input>
+ <port id="0">
+ <dim>8</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="1" precision="FP16">
+ <dim>8</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="351" name="bottleneck1_4/dim_red/conv/fq_weights_1/zero_point1279722296" type="Const" version="opset1">
+ <data element_type="f16" offset="4308" shape="8,1,1,1" size="16"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="352" name="bottleneck1_4/dim_red/conv/fq_weights_1/minus_zp" type="Subtract" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>8</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="1">
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>8</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="353" name="bottleneck1_4/dim_red/conv/fq_weights_1/scale1279221831" type="Const" version="opset1">
+ <data element_type="f16" offset="4324" shape="8,1,1,1" size="16"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="354" name="bottleneck1_4/dim_red/conv/fq_weights_1/mulpiply_by_scale" type="Multiply" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>8</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="1">
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>8</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="355" name="bottleneck1_4/dim_red/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1,1" pads_begin="0,0" pads_end="0,0" strides="1,1"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ <port id="1">
+ <dim>8</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="356" name="data_add_237292373481520922" type="Const" version="opset1">
+ <data element_type="f16" offset="4340" shape="1,8,1,1" size="16"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="357" name="bottleneck1_4/dim_red/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck1_4/dim_red/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="358" name="bottleneck1_4/dim_red/fn/weights3094840595817" type="Const" version="opset1">
+ <data element_type="f32" offset="1576" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="359" name="bottleneck1_4/dim_red/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck1_4/dim_red/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="360" name="bottleneck1_4/inner/dw1/conv/fq_input_0" type="FakeQuantize" version="opset1">
+ <data auto_broadcast="numpy" levels="256"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="2">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="3">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="4">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="5" precision="FP16">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="361" name="16815/value1681722659" type="Const" version="opset1">
+ <data element_type="i64" offset="1580" shape="5" size="40"/>
+ <output>
+ <port id="0" precision="I64">
+ <dim>5</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="362" name="bottleneck1_4/inner/dw1/bn/mean/Fused_Mul__copy81910112/quantized1350421549" type="Const" version="opset1">
+ <data element_type="i8" offset="4356" shape="8,1,3,3" size="72"/>
+ <output>
+ <port id="0" precision="I8">
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="363" name="bottleneck1_4/inner/dw1/bn/mean/Fused_Mul__copy81910112/quantized/to_f16" type="Convert" version="opset1">
+ <data destination_type="f16"/>
+ <input>
+ <port id="0">
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </input>
+ <output>
+ <port id="1" precision="FP16">
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="364" name="bottleneck1_4/inner/dw1/conv/fq_weights_1/zero_point1351721285" type="Const" version="opset1">
+ <data element_type="f16" offset="4428" shape="8,1,1,1" size="16"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="365" name="bottleneck1_4/inner/dw1/conv/fq_weights_1/minus_zp" type="Subtract" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ <port id="1">
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="366" name="bottleneck1_4/inner/dw1/conv/fq_weights_1/scale1351221033" type="Const" version="opset1">
+ <data element_type="f16" offset="4444" shape="8,1,1,1" size="16"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="367" name="bottleneck1_4/inner/dw1/conv/fq_weights_1/mulpiply_by_scale" type="Multiply" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ <port id="1">
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="368" name="16815" type="Reshape" version="opset1">
+ <data special_zero="true"/>
+ <input>
+ <port id="0">
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ <port id="1">
+ <dim>5</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="369" name="bottleneck1_4/inner/dw1/conv" type="GroupConvolution" version="opset1">
+ <data auto_pad="explicit" dilations="1,1" pads_begin="1,1" pads_end="1,1" strides="1,1"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ <port id="1">
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="370" name="data_add_237372374282122335" type="Const" version="opset1">
+ <data element_type="f16" offset="4460" shape="1,8,1,1" size="16"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="371" name="bottleneck1_4/inner/dw1/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck1_4/inner/dw1/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="372" name="bottleneck1_4/inner/dw1/fn/weights3104439704823" type="Const" version="opset1">
+ <data element_type="f32" offset="1576" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="373" name="bottleneck1_4/inner/dw1/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck1_4/inner/dw1/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="374" name="bottleneck1_4/dim_inc/conv/fq_input_0" type="FakeQuantize" version="opset1">
+ <data auto_broadcast="numpy" levels="256"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ <port id="1"/>
+ <port id="2"/>
+ <port id="3"/>
+ <port id="4"/>
+ </input>
+ <output>
+ <port id="5" precision="FP16">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="375" name="bottleneck1_4/dim_inc/bn/mean/Fused_Mul__copy82510115/quantized1232820436" type="Const" version="opset1">
+ <data element_type="i8" offset="4476" shape="32,8,1,1" size="256"/>
+ <output>
+ <port id="0" precision="I8">
+ <dim>32</dim>
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="376" name="bottleneck1_4/dim_inc/bn/mean/Fused_Mul__copy82510115/quantized/to_f16" type="Convert" version="opset1">
+ <data destination_type="f16"/>
+ <input>
+ <port id="0">
+ <dim>32</dim>
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="1" precision="FP16">
+ <dim>32</dim>
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="377" name="bottleneck1_4/dim_inc/conv/fq_weights_1/zero_point1234119761" type="Const" version="opset1">
+ <data element_type="f16" offset="4732" shape="32,1,1,1" size="64"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="378" name="bottleneck1_4/dim_inc/conv/fq_weights_1/minus_zp" type="Subtract" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>32</dim>
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="1">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>32</dim>
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="379" name="bottleneck1_4/dim_inc/conv/fq_weights_1/scale1233622032" type="Const" version="opset1">
+ <data element_type="f16" offset="4796" shape="32,1,1,1" size="64"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="380" name="bottleneck1_4/dim_inc/conv/fq_weights_1/mulpiply_by_scale" type="Multiply" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>32</dim>
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="1">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>32</dim>
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="381" name="bottleneck1_4/dim_inc/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1,1" pads_begin="0,0" pads_end="0,0" strides="1,1"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ <port id="1">
+ <dim>32</dim>
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="382" name="data_add_237452375082720163" type="Const" version="opset1">
+ <data element_type="f16" offset="4860" shape="1,32,1,1" size="64"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="383" name="bottleneck1_4/dim_inc/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck1_4/dim_inc/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="384" name="bottleneck1_4/add/fq_input_1" type="FakeQuantize" version="opset1">
+ <data auto_broadcast="numpy" levels="256"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ <port id="1"/>
+ <port id="2"/>
+ <port id="3"/>
+ <port id="4"/>
+ </input>
+ <output>
+ <port id="5" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="385" name="bottleneck1_4/add" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck1_4/add" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="386" name="bottleneck1_4/fn/weights3103640535830" type="Const" version="opset1">
+ <data element_type="f32" offset="1576" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="387" name="bottleneck1_4/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck1_4/add" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="388" name="bottleneck2_0/dim_red/conv/fq_input_0" type="FakeQuantize" version="opset1">
+ <data auto_broadcast="numpy" levels="256"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ <port id="1"/>
+ <port id="2"/>
+ <port id="3"/>
+ <port id="4"/>
+ </input>
+ <output>
+ <port id="5" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="389" name="bottleneck2_0/skip/pooling" type="MaxPool" version="opset1">
+ <data auto_pad="explicit" kernel="2,2" pads_begin="0,0" pads_end="0,0" rounding_type="ceil" strides="2,2"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ </input>
+ <output>
+ <port id="1" names="bottleneck2_0/skip/pooling" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="390" name="bottleneck2_0/skip/bn/mean/Fused_Mul__copy83310118/quantized1230422248" type="Const" version="opset1">
+ <data element_type="i8" offset="4924" shape="64,32,1,1" size="2048"/>
+ <output>
+ <port id="0" precision="I8">
+ <dim>64</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="391" name="bottleneck2_0/skip/bn/mean/Fused_Mul__copy83310118/quantized/to_f16" type="Convert" version="opset1">
+ <data destination_type="f16"/>
+ <input>
+ <port id="0">
+ <dim>64</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="1" precision="FP16">
+ <dim>64</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="392" name="bottleneck2_0/skip/conv/fq_weights_1/zero_point1231721753" type="Const" version="opset1">
+ <data element_type="f16" offset="6972" shape="64,1,1,1" size="128"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="393" name="bottleneck2_0/skip/conv/fq_weights_1/minus_zp" type="Subtract" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>64</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="1">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>64</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="394" name="bottleneck2_0/skip/conv/fq_weights_1/scale1231221261" type="Const" version="opset1">
+ <data element_type="f16" offset="7100" shape="64,1,1,1" size="128"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="395" name="bottleneck2_0/skip/conv/fq_weights_1/mulpiply_by_scale" type="Multiply" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>64</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="1">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>64</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="396" name="bottleneck2_0/skip/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1,1" pads_begin="0,0" pads_end="0,0" strides="1,1"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1">
+ <dim>64</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="397" name="data_add_237532375883522425" type="Const" version="opset1">
+ <data element_type="f16" offset="7228" shape="1,64,1,1" size="128"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="398" name="bottleneck2_0/skip/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck2_0/skip/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="399" name="bottleneck2_0/add/fq_input_0" type="FakeQuantize" version="opset1">
+ <data auto_broadcast="numpy" levels="256"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1"/>
+ <port id="2"/>
+ <port id="3"/>
+ <port id="4"/>
+ </input>
+ <output>
+ <port id="5" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="400" name="2734273822632" type="Const" version="opset1">
+ <data element_type="f16" offset="7356" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="401" name="2735273921387" type="Const" version="opset1">
+ <data element_type="f16" offset="7358" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="402" name="2736274020394" type="Const" version="opset1">
+ <data element_type="f16" offset="7356" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="403" name="2737274121651" type="Const" version="opset1">
+ <data element_type="f16" offset="7358" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="404" name="5244524821498" type="Const" version="opset1">
+ <data element_type="f16" offset="7360" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="405" name="5245524919710" type="Const" version="opset1">
+ <data element_type="f16" offset="7362" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="406" name="5246525019878" type="Const" version="opset1">
+ <data element_type="f16" offset="7360" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="407" name="5247525121837" type="Const" version="opset1">
+ <data element_type="f16" offset="7362" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="408" name="2564256820559" type="Const" version="opset1">
+ <data element_type="f16" offset="7364" shape="1,16,1,1" size="32"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="409" name="2565256922602" type="Const" version="opset1">
+ <data element_type="f16" offset="7396" shape="1,16,1,1" size="32"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="410" name="2566257020511" type="Const" version="opset1">
+ <data element_type="f16" offset="7364" shape="1,16,1,1" size="32"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="411" name="2567257122908" type="Const" version="opset1">
+ <data element_type="f16" offset="7396" shape="1,16,1,1" size="32"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="412" name="bottleneck2_0/dim_red/bn/mean/Fused_Mul__copy83710120/quantized1213621963" type="Const" version="opset1">
+ <data element_type="i8" offset="7428" shape="16,32,1,1" size="512"/>
+ <output>
+ <port id="0" precision="I8">
+ <dim>16</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="413" name="bottleneck2_0/dim_red/bn/mean/Fused_Mul__copy83710120/quantized/to_f16" type="Convert" version="opset1">
+ <data destination_type="f16"/>
+ <input>
+ <port id="0">
+ <dim>16</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="1" precision="FP16">
+ <dim>16</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="414" name="bottleneck2_0/dim_red/conv/fq_weights_1/zero_point1214921390" type="Const" version="opset1">
+ <data element_type="f16" offset="7940" shape="16,1,1,1" size="32"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="415" name="bottleneck2_0/dim_red/conv/fq_weights_1/minus_zp" type="Subtract" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>16</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="1">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>16</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="416" name="bottleneck2_0/dim_red/conv/fq_weights_1/scale1214419473" type="Const" version="opset1">
+ <data element_type="f16" offset="7972" shape="16,1,1,1" size="32"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="417" name="bottleneck2_0/dim_red/conv/fq_weights_1/mulpiply_by_scale" type="Multiply" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>16</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="1">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>16</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="418" name="bottleneck2_0/dim_red/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1,1" pads_begin="0,0" pads_end="0,0" strides="1,1"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ <port id="1">
+ <dim>16</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="419" name="data_add_237612376683920037" type="Const" version="opset1">
+ <data element_type="f16" offset="8004" shape="1,16,1,1" size="32"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="420" name="bottleneck2_0/dim_red/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck2_0/dim_red/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="421" name="bottleneck2_0/dim_red/fn/weights3109240148841" type="Const" version="opset1">
+ <data element_type="f32" offset="1576" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="422" name="bottleneck2_0/dim_red/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck2_0/dim_red/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="423" name="bottleneck2_0/inner/dw1/conv/fq_input_0" type="FakeQuantize" version="opset1">
+ <data auto_broadcast="numpy" levels="256"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="2">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="3">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="4">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="5" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="424" name="16799/value1680122737" type="Const" version="opset1">
+ <data element_type="i64" offset="8036" shape="5" size="40"/>
+ <output>
+ <port id="0" precision="I64">
+ <dim>5</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="425" name="bottleneck2_0/inner/dw1/bn/mean/Fused_Mul__copy84310123/quantized1314419509" type="Const" version="opset1">
+ <data element_type="i8" offset="8076" shape="16,1,3,3" size="144"/>
+ <output>
+ <port id="0" precision="I8">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="426" name="bottleneck2_0/inner/dw1/bn/mean/Fused_Mul__copy84310123/quantized/to_f16" type="Convert" version="opset1">
+ <data destination_type="f16"/>
+ <input>
+ <port id="0">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </input>
+ <output>
+ <port id="1" precision="FP16">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="427" name="bottleneck2_0/inner/dw1/conv/fq_weights_1/zero_point1315720619" type="Const" version="opset1">
+ <data element_type="f16" offset="8220" shape="16,1,1,1" size="32"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="428" name="bottleneck2_0/inner/dw1/conv/fq_weights_1/minus_zp" type="Subtract" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ <port id="1">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="429" name="bottleneck2_0/inner/dw1/conv/fq_weights_1/scale1315221030" type="Const" version="opset1">
+ <data element_type="f16" offset="8252" shape="16,1,1,1" size="32"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="430" name="bottleneck2_0/inner/dw1/conv/fq_weights_1/mulpiply_by_scale" type="Multiply" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ <port id="1">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="431" name="16799" type="Reshape" version="opset1">
+ <data special_zero="true"/>
+ <input>
+ <port id="0">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ <port id="1">
+ <dim>5</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="432" name="bottleneck2_0/inner/dw1/conv" type="GroupConvolution" version="opset1">
+ <data auto_pad="explicit" dilations="1,1" pads_begin="1,1" pads_end="1,1" strides="2,2"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ <port id="1">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="433" name="data_add_237692377484521879" type="Const" version="opset1">
+ <data element_type="f16" offset="8284" shape="1,16,1,1" size="32"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="434" name="bottleneck2_0/inner/dw1/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck2_0/inner/dw1/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="435" name="bottleneck2_0/inner/dw1/fn/weights3088440433847" type="Const" version="opset1">
+ <data element_type="f32" offset="1576" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="436" name="bottleneck2_0/inner/dw1/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck2_0/inner/dw1/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="437" name="bottleneck2_0/dim_inc/conv/fq_input_0" type="FakeQuantize" version="opset1">
+ <data auto_broadcast="numpy" levels="256"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1"/>
+ <port id="2"/>
+ <port id="3"/>
+ <port id="4"/>
+ </input>
+ <output>
+ <port id="5" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="438" name="bottleneck2_0/dim_inc/bn/mean/Fused_Mul__copy84910126/quantized1172822140" type="Const" version="opset1">
+ <data element_type="i8" offset="8316" shape="64,16,1,1" size="1024"/>
+ <output>
+ <port id="0" precision="I8">
+ <dim>64</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="439" name="bottleneck2_0/dim_inc/bn/mean/Fused_Mul__copy84910126/quantized/to_f16" type="Convert" version="opset1">
+ <data destination_type="f16"/>
+ <input>
+ <port id="0">
+ <dim>64</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="1" precision="FP16">
+ <dim>64</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="440" name="bottleneck2_0/dim_inc/conv/fq_weights_1/zero_point1174120334" type="Const" version="opset1">
+ <data element_type="f16" offset="9340" shape="64,1,1,1" size="128"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="441" name="bottleneck2_0/dim_inc/conv/fq_weights_1/minus_zp" type="Subtract" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>64</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="1">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>64</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="442" name="bottleneck2_0/dim_inc/conv/fq_weights_1/scale1173620184" type="Const" version="opset1">
+ <data element_type="f16" offset="9468" shape="64,1,1,1" size="128"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="443" name="bottleneck2_0/dim_inc/conv/fq_weights_1/mulpiply_by_scale" type="Multiply" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>64</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="1">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>64</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="444" name="bottleneck2_0/dim_inc/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1,1" pads_begin="0,0" pads_end="0,0" strides="1,1"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1">
+ <dim>64</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="445" name="data_add_237772378285120616" type="Const" version="opset1">
+ <data element_type="f16" offset="9596" shape="1,64,1,1" size="128"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="446" name="bottleneck2_0/dim_inc/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck2_0/dim_inc/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="447" name="bottleneck2_0/add/fq_input_1" type="FakeQuantize" version="opset1">
+ <data auto_broadcast="numpy" levels="256"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1"/>
+ <port id="2"/>
+ <port id="3"/>
+ <port id="4"/>
+ </input>
+ <output>
+ <port id="5" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="448" name="bottleneck2_0/add" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck2_0/add" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="449" name="bottleneck2_0/fn/weights3105639815854" type="Const" version="opset1">
+ <data element_type="f32" offset="1576" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="450" name="bottleneck2_0/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck2_0/add" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="451" name="bottleneck2_1/add/fq_input_0" type="FakeQuantize" version="opset1">
+ <data auto_broadcast="numpy" levels="256"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1"/>
+ <port id="2"/>
+ <port id="3"/>
+ <port id="4"/>
+ </input>
+ <output>
+ <port id="5" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="452" name="5314531820286" type="Const" version="opset1">
+ <data element_type="f16" offset="9724" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="453" name="5315531919620" type="Const" version="opset1">
+ <data element_type="f16" offset="9726" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="454" name="5316532021444" type="Const" version="opset1">
+ <data element_type="f16" offset="9724" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="455" name="5317532122701" type="Const" version="opset1">
+ <data element_type="f16" offset="9726" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="456" name="5424542822068" type="Const" version="opset1">
+ <data element_type="f16" offset="9728" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="457" name="5425542920760" type="Const" version="opset1">
+ <data element_type="f16" offset="9730" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="458" name="5426543021957" type="Const" version="opset1">
+ <data element_type="f16" offset="9728" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="459" name="5427543119944" type="Const" version="opset1">
+ <data element_type="f16" offset="9730" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="460" name="3864386820688" type="Const" version="opset1">
+ <data element_type="f16" offset="9732" shape="1,16,1,1" size="32"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="461" name="3865386919656" type="Const" version="opset1">
+ <data element_type="f16" offset="9764" shape="1,16,1,1" size="32"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="462" name="3866387022311" type="Const" version="opset1">
+ <data element_type="f16" offset="9732" shape="1,16,1,1" size="32"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="463" name="3867387121090" type="Const" version="opset1">
+ <data element_type="f16" offset="9764" shape="1,16,1,1" size="32"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="464" name="bottleneck2_1/dim_red/bn/mean/Fused_Mul__copy85610129/quantized1324022005" type="Const" version="opset1">
+ <data element_type="i8" offset="9796" shape="16,64,1,1" size="1024"/>
+ <output>
+ <port id="0" precision="I8">
+ <dim>16</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="465" name="bottleneck2_1/dim_red/bn/mean/Fused_Mul__copy85610129/quantized/to_f16" type="Convert" version="opset1">
+ <data destination_type="f16"/>
+ <input>
+ <port id="0">
+ <dim>16</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="1" precision="FP16">
+ <dim>16</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="466" name="bottleneck2_1/dim_red/conv/fq_weights_1/zero_point1325321120" type="Const" version="opset1">
+ <data element_type="f16" offset="10820" shape="16,1,1,1" size="32"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="467" name="bottleneck2_1/dim_red/conv/fq_weights_1/minus_zp" type="Subtract" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>16</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="1">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>16</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="468" name="bottleneck2_1/dim_red/conv/fq_weights_1/scale1324822167" type="Const" version="opset1">
+ <data element_type="f16" offset="10852" shape="16,1,1,1" size="32"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="469" name="bottleneck2_1/dim_red/conv/fq_weights_1/mulpiply_by_scale" type="Multiply" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>16</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="1">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>16</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="470" name="bottleneck2_1/dim_red/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1,1" pads_begin="0,0" pads_end="0,0" strides="1,1"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1">
+ <dim>16</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="471" name="data_add_237852379085821495" type="Const" version="opset1">
+ <data element_type="f16" offset="10884" shape="1,16,1,1" size="32"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="472" name="bottleneck2_1/dim_red/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck2_1/dim_red/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="473" name="bottleneck2_1/dim_red/fn/weights3104840109860" type="Const" version="opset1">
+ <data element_type="f32" offset="1576" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="474" name="bottleneck2_1/dim_red/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck2_1/dim_red/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="475" name="bottleneck2_1/inner/dw1/conv/fq_input_0" type="FakeQuantize" version="opset1">
+ <data auto_broadcast="numpy" levels="256"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="2">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="3">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="4">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="5" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="476" name="16859/value1686121339" type="Const" version="opset1">
+ <data element_type="i64" offset="8036" shape="5" size="40"/>
+ <output>
+ <port id="0" precision="I64">
+ <dim>5</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="477" name="bottleneck2_1/inner/dw1/bn/mean/Fused_Mul__copy86210132/quantized1391221243" type="Const" version="opset1">
+ <data element_type="i8" offset="10916" shape="16,1,3,3" size="144"/>
+ <output>
+ <port id="0" precision="I8">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="478" name="bottleneck2_1/inner/dw1/bn/mean/Fused_Mul__copy86210132/quantized/to_f16" type="Convert" version="opset1">
+ <data destination_type="f16"/>
+ <input>
+ <port id="0">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </input>
+ <output>
+ <port id="1" precision="FP16">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="479" name="bottleneck2_1/inner/dw1/conv/fq_weights_1/zero_point1392520535" type="Const" version="opset1">
+ <data element_type="f16" offset="11060" shape="16,1,1,1" size="32"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="480" name="bottleneck2_1/inner/dw1/conv/fq_weights_1/minus_zp" type="Subtract" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ <port id="1">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="481" name="bottleneck2_1/inner/dw1/conv/fq_weights_1/scale1392021039" type="Const" version="opset1">
+ <data element_type="f16" offset="11092" shape="16,1,1,1" size="32"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="482" name="bottleneck2_1/inner/dw1/conv/fq_weights_1/mulpiply_by_scale" type="Multiply" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ <port id="1">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="483" name="16859" type="Reshape" version="opset1">
+ <data special_zero="true"/>
+ <input>
+ <port id="0">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ <port id="1">
+ <dim>5</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="484" name="bottleneck2_1/inner/dw1/conv" type="GroupConvolution" version="opset1">
+ <data auto_pad="explicit" dilations="1,1" pads_begin="1,1" pads_end="1,1" strides="1,1"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="485" name="data_add_237932379886422692" type="Const" version="opset1">
+ <data element_type="f16" offset="11124" shape="1,16,1,1" size="32"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="486" name="bottleneck2_1/inner/dw1/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck2_1/inner/dw1/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="487" name="bottleneck2_1/inner/dw1/fn/weights3086040310866" type="Const" version="opset1">
+ <data element_type="f32" offset="1576" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="488" name="bottleneck2_1/inner/dw1/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck2_1/inner/dw1/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="489" name="bottleneck2_1/dim_inc/conv/fq_input_0" type="FakeQuantize" version="opset1">
+ <data auto_broadcast="numpy" levels="256"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1"/>
+ <port id="2"/>
+ <port id="3"/>
+ <port id="4"/>
+ </input>
+ <output>
+ <port id="5" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="490" name="bottleneck2_1/dim_inc/bn/mean/Fused_Mul__copy86810135/quantized1360021273" type="Const" version="opset1">
+ <data element_type="i8" offset="11156" shape="64,16,1,1" size="1024"/>
+ <output>
+ <port id="0" precision="I8">
+ <dim>64</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="491" name="bottleneck2_1/dim_inc/bn/mean/Fused_Mul__copy86810135/quantized/to_f16" type="Convert" version="opset1">
+ <data destination_type="f16"/>
+ <input>
+ <port id="0">
+ <dim>64</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="1" precision="FP16">
+ <dim>64</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="492" name="bottleneck2_1/dim_inc/conv/fq_weights_1/zero_point1361322104" type="Const" version="opset1">
+ <data element_type="f16" offset="12180" shape="64,1,1,1" size="128"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="493" name="bottleneck2_1/dim_inc/conv/fq_weights_1/minus_zp" type="Subtract" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>64</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="1">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>64</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="494" name="bottleneck2_1/dim_inc/conv/fq_weights_1/scale1360821006" type="Const" version="opset1">
+ <data element_type="f16" offset="12308" shape="64,1,1,1" size="128"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="495" name="bottleneck2_1/dim_inc/conv/fq_weights_1/mulpiply_by_scale" type="Multiply" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>64</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="1">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>64</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="496" name="bottleneck2_1/dim_inc/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1,1" pads_begin="0,0" pads_end="0,0" strides="1,1"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1">
+ <dim>64</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="497" name="data_add_238012380687019767" type="Const" version="opset1">
+ <data element_type="f16" offset="12436" shape="1,64,1,1" size="128"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="498" name="bottleneck2_1/dim_inc/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck2_1/dim_inc/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="499" name="bottleneck2_1/add/fq_input_1" type="FakeQuantize" version="opset1">
+ <data auto_broadcast="numpy" levels="256"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1"/>
+ <port id="2"/>
+ <port id="3"/>
+ <port id="4"/>
+ </input>
+ <output>
+ <port id="5" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="500" name="bottleneck2_1/add" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck2_1/add" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="501" name="bottleneck2_1/fn/weights3117640334873" type="Const" version="opset1">
+ <data element_type="f32" offset="1576" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="502" name="bottleneck2_1/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck2_1/add" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="503" name="bottleneck2_2/add/fq_input_0" type="FakeQuantize" version="opset1">
+ <data auto_broadcast="numpy" levels="256"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1"/>
+ <port id="2"/>
+ <port id="3"/>
+ <port id="4"/>
+ </input>
+ <output>
+ <port id="5" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="504" name="4974497821771" type="Const" version="opset1">
+ <data element_type="f16" offset="12564" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="505" name="4975497920907" type="Const" version="opset1">
+ <data element_type="f16" offset="12566" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="506" name="4976498020433" type="Const" version="opset1">
+ <data element_type="f16" offset="12564" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="507" name="4977498120295" type="Const" version="opset1">
+ <data element_type="f16" offset="12566" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="508" name="3244324821267" type="Const" version="opset1">
+ <data element_type="f16" offset="12568" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="509" name="3245324920004" type="Const" version="opset1">
+ <data element_type="f16" offset="12570" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="510" name="3246325021405" type="Const" version="opset1">
+ <data element_type="f16" offset="12568" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="511" name="3247325121666" type="Const" version="opset1">
+ <data element_type="f16" offset="12570" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="512" name="4724472819536" type="Const" version="opset1">
+ <data element_type="f16" offset="12572" shape="1,16,1,1" size="32"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="513" name="4725472919467" type="Const" version="opset1">
+ <data element_type="f16" offset="12604" shape="1,16,1,1" size="32"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="514" name="4726473022512" type="Const" version="opset1">
+ <data element_type="f16" offset="12572" shape="1,16,1,1" size="32"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="515" name="4727473120211" type="Const" version="opset1">
+ <data element_type="f16" offset="12604" shape="1,16,1,1" size="32"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="516" name="bottleneck2_2/dim_red/bn/mean/Fused_Mul__copy87510138/quantized1410422479" type="Const" version="opset1">
+ <data element_type="i8" offset="12636" shape="16,64,1,1" size="1024"/>
+ <output>
+ <port id="0" precision="I8">
+ <dim>16</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="517" name="bottleneck2_2/dim_red/bn/mean/Fused_Mul__copy87510138/quantized/to_f16" type="Convert" version="opset1">
+ <data destination_type="f16"/>
+ <input>
+ <port id="0">
+ <dim>16</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="1" precision="FP16">
+ <dim>16</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="518" name="bottleneck2_2/dim_red/conv/fq_weights_1/zero_point1411722911" type="Const" version="opset1">
+ <data element_type="f16" offset="13660" shape="16,1,1,1" size="32"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="519" name="bottleneck2_2/dim_red/conv/fq_weights_1/minus_zp" type="Subtract" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>16</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="1">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>16</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="520" name="bottleneck2_2/dim_red/conv/fq_weights_1/scale1411222305" type="Const" version="opset1">
+ <data element_type="f16" offset="13692" shape="16,1,1,1" size="32"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="521" name="bottleneck2_2/dim_red/conv/fq_weights_1/mulpiply_by_scale" type="Multiply" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>16</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="1">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>16</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="522" name="bottleneck2_2/dim_red/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1,1" pads_begin="0,0" pads_end="0,0" strides="1,1"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1">
+ <dim>16</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="523" name="data_add_238092381487722668" type="Const" version="opset1">
+ <data element_type="f16" offset="13724" shape="1,16,1,1" size="32"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="524" name="bottleneck2_2/dim_red/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck2_2/dim_red/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="525" name="bottleneck2_2/dim_red/fn/weights3090840547879" type="Const" version="opset1">
+ <data element_type="f32" offset="1576" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="526" name="bottleneck2_2/dim_red/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck2_2/dim_red/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="527" name="bottleneck2_2/inner/dw1/conv/fq_input_0" type="FakeQuantize" version="opset1">
+ <data auto_broadcast="numpy" levels="256"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="2">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="3">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="4">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="5" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="528" name="16895/value1689719797" type="Const" version="opset1">
+ <data element_type="i64" offset="8036" shape="5" size="40"/>
+ <output>
+ <port id="0" precision="I64">
+ <dim>5</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="529" name="bottleneck2_2/inner/dw1/bn/mean/Fused_Mul__copy88110141/quantized1158420547" type="Const" version="opset1">
+ <data element_type="i8" offset="13756" shape="16,1,3,3" size="144"/>
+ <output>
+ <port id="0" precision="I8">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="530" name="bottleneck2_2/inner/dw1/bn/mean/Fused_Mul__copy88110141/quantized/to_f16" type="Convert" version="opset1">
+ <data destination_type="f16"/>
+ <input>
+ <port id="0">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </input>
+ <output>
+ <port id="1" precision="FP16">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="531" name="bottleneck2_2/inner/dw1/conv/fq_weights_1/zero_point1159721558" type="Const" version="opset1">
+ <data element_type="f16" offset="13900" shape="16,1,1,1" size="32"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="532" name="bottleneck2_2/inner/dw1/conv/fq_weights_1/minus_zp" type="Subtract" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ <port id="1">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="533" name="bottleneck2_2/inner/dw1/conv/fq_weights_1/scale1159219389" type="Const" version="opset1">
+ <data element_type="f16" offset="13932" shape="16,1,1,1" size="32"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="534" name="bottleneck2_2/inner/dw1/conv/fq_weights_1/mulpiply_by_scale" type="Multiply" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ <port id="1">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="535" name="16895" type="Reshape" version="opset1">
+ <data special_zero="true"/>
+ <input>
+ <port id="0">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ <port id="1">
+ <dim>5</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="536" name="bottleneck2_2/inner/dw1/conv" type="GroupConvolution" version="opset1">
+ <data auto_pad="explicit" dilations="1,1" pads_begin="1,1" pads_end="1,1" strides="1,1"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="537" name="data_add_238172382288322365" type="Const" version="opset1">
+ <data element_type="f16" offset="13964" shape="1,16,1,1" size="32"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="538" name="bottleneck2_2/inner/dw1/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck2_2/inner/dw1/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="539" name="bottleneck2_2/inner/dw1/fn/weights3118440412885" type="Const" version="opset1">
+ <data element_type="f32" offset="1576" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="540" name="bottleneck2_2/inner/dw1/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck2_2/inner/dw1/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="541" name="bottleneck2_2/dim_inc/conv/fq_input_0" type="FakeQuantize" version="opset1">
+ <data auto_broadcast="numpy" levels="256"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1"/>
+ <port id="2"/>
+ <port id="3"/>
+ <port id="4"/>
+ </input>
+ <output>
+ <port id="5" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="542" name="bottleneck2_2/dim_inc/bn/mean/Fused_Mul__copy88710144/quantized1206419590" type="Const" version="opset1">
+ <data element_type="i8" offset="13996" shape="64,16,1,1" size="1024"/>
+ <output>
+ <port id="0" precision="I8">
+ <dim>64</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="543" name="bottleneck2_2/dim_inc/bn/mean/Fused_Mul__copy88710144/quantized/to_f16" type="Convert" version="opset1">
+ <data destination_type="f16"/>
+ <input>
+ <port id="0">
+ <dim>64</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="1" precision="FP16">
+ <dim>64</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="544" name="bottleneck2_2/dim_inc/conv/fq_weights_1/zero_point1207720709" type="Const" version="opset1">
+ <data element_type="f16" offset="15020" shape="64,1,1,1" size="128"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="545" name="bottleneck2_2/dim_inc/conv/fq_weights_1/minus_zp" type="Subtract" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>64</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="1">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>64</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="546" name="bottleneck2_2/dim_inc/conv/fq_weights_1/scale1207219722" type="Const" version="opset1">
+ <data element_type="f16" offset="15148" shape="64,1,1,1" size="128"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="547" name="bottleneck2_2/dim_inc/conv/fq_weights_1/mulpiply_by_scale" type="Multiply" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>64</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="1">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>64</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="548" name="bottleneck2_2/dim_inc/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1,1" pads_begin="0,0" pads_end="0,0" strides="1,1"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1">
+ <dim>64</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="549" name="data_add_238252383088922611" type="Const" version="opset1">
+ <data element_type="f16" offset="15276" shape="1,64,1,1" size="128"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="550" name="bottleneck2_2/dim_inc/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck2_2/dim_inc/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="551" name="bottleneck2_2/add/fq_input_1" type="FakeQuantize" version="opset1">
+ <data auto_broadcast="numpy" levels="256"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1"/>
+ <port id="2"/>
+ <port id="3"/>
+ <port id="4"/>
+ </input>
+ <output>
+ <port id="5" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="552" name="bottleneck2_2/add" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck2_2/add" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="553" name="bottleneck2_2/fn/weights3106040232892" type="Const" version="opset1">
+ <data element_type="f32" offset="1576" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="554" name="bottleneck2_2/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck2_2/add" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="555" name="bottleneck2_3/add/fq_input_0" type="FakeQuantize" version="opset1">
+ <data auto_broadcast="numpy" levels="256"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1"/>
+ <port id="2"/>
+ <port id="3"/>
+ <port id="4"/>
+ </input>
+ <output>
+ <port id="5" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="556" name="5414541820298" type="Const" version="opset1">
+ <data element_type="f16" offset="15404" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="557" name="5415541922815" type="Const" version="opset1">
+ <data element_type="f16" offset="15406" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="558" name="5416542021462" type="Const" version="opset1">
+ <data element_type="f16" offset="15404" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="559" name="5417542122590" type="Const" version="opset1">
+ <data element_type="f16" offset="15406" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="560" name="4564456822848" type="Const" version="opset1">
+ <data element_type="f16" offset="15408" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="561" name="4565456920025" type="Const" version="opset1">
+ <data element_type="f16" offset="15410" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="562" name="4566457020655" type="Const" version="opset1">
+ <data element_type="f16" offset="15408" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="563" name="4567457121222" type="Const" version="opset1">
+ <data element_type="f16" offset="15410" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="564" name="4104410820580" type="Const" version="opset1">
+ <data element_type="f16" offset="15412" shape="1,16,1,1" size="32"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="565" name="4105410921249" type="Const" version="opset1">
+ <data element_type="f16" offset="15444" shape="1,16,1,1" size="32"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="566" name="4106411021675" type="Const" version="opset1">
+ <data element_type="f16" offset="15412" shape="1,16,1,1" size="32"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="567" name="4107411121321" type="Const" version="opset1">
+ <data element_type="f16" offset="15444" shape="1,16,1,1" size="32"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="568" name="bottleneck2_3/dim_red/bn/mean/Fused_Mul__copy89410147/quantized1177619557" type="Const" version="opset1">
+ <data element_type="i8" offset="15476" shape="16,64,1,1" size="1024"/>
+ <output>
+ <port id="0" precision="I8">
+ <dim>16</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="569" name="bottleneck2_3/dim_red/bn/mean/Fused_Mul__copy89410147/quantized/to_f16" type="Convert" version="opset1">
+ <data destination_type="f16"/>
+ <input>
+ <port id="0">
+ <dim>16</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="1" precision="FP16">
+ <dim>16</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="570" name="bottleneck2_3/dim_red/conv/fq_weights_1/zero_point1178920415" type="Const" version="opset1">
+ <data element_type="f16" offset="16500" shape="16,1,1,1" size="32"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="571" name="bottleneck2_3/dim_red/conv/fq_weights_1/minus_zp" type="Subtract" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>16</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="1">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>16</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="572" name="bottleneck2_3/dim_red/conv/fq_weights_1/scale1178419392" type="Const" version="opset1">
+ <data element_type="f16" offset="16532" shape="16,1,1,1" size="32"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="573" name="bottleneck2_3/dim_red/conv/fq_weights_1/mulpiply_by_scale" type="Multiply" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>16</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="1">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>16</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="574" name="bottleneck2_3/dim_red/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1,1" pads_begin="0,0" pads_end="0,0" strides="1,1"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1">
+ <dim>16</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="575" name="data_add_238332383889619500" type="Const" version="opset1">
+ <data element_type="f16" offset="16564" shape="1,16,1,1" size="32"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="576" name="bottleneck2_3/dim_red/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck2_3/dim_red/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="577" name="bottleneck2_3/dim_red/fn/weights3109640157898" type="Const" version="opset1">
+ <data element_type="f32" offset="1576" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="578" name="bottleneck2_3/dim_red/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck2_3/dim_red/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="579" name="bottleneck2_3/inner/dw1/conv/fq_input_0" type="FakeQuantize" version="opset1">
+ <data auto_broadcast="numpy" levels="256"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="2">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="3">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="4">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="5" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="580" name="16871/value1687322308" type="Const" version="opset1">
+ <data element_type="i64" offset="8036" shape="5" size="40"/>
+ <output>
+ <port id="0" precision="I64">
+ <dim>5</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="581" name="bottleneck2_3/inner/dw1/bn/mean/Fused_Mul__copy90010150/quantized1208821009" type="Const" version="opset1">
+ <data element_type="i8" offset="16596" shape="16,1,3,3" size="144"/>
+ <output>
+ <port id="0" precision="I8">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="582" name="bottleneck2_3/inner/dw1/bn/mean/Fused_Mul__copy90010150/quantized/to_f16" type="Convert" version="opset1">
+ <data destination_type="f16"/>
+ <input>
+ <port id="0">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </input>
+ <output>
+ <port id="1" precision="FP16">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="583" name="bottleneck2_3/inner/dw1/conv/fq_weights_1/zero_point1210121105" type="Const" version="opset1">
+ <data element_type="f16" offset="16740" shape="16,1,1,1" size="32"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="584" name="bottleneck2_3/inner/dw1/conv/fq_weights_1/minus_zp" type="Subtract" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ <port id="1">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="585" name="bottleneck2_3/inner/dw1/conv/fq_weights_1/scale1209619950" type="Const" version="opset1">
+ <data element_type="f16" offset="16772" shape="16,1,1,1" size="32"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="586" name="bottleneck2_3/inner/dw1/conv/fq_weights_1/mulpiply_by_scale" type="Multiply" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ <port id="1">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="587" name="16871" type="Reshape" version="opset1">
+ <data special_zero="true"/>
+ <input>
+ <port id="0">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ <port id="1">
+ <dim>5</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="588" name="bottleneck2_3/inner/dw1/conv" type="GroupConvolution" version="opset1">
+ <data auto_pad="explicit" dilations="1,1" pads_begin="1,1" pads_end="1,1" strides="1,1"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="589" name="data_add_238412384690221054" type="Const" version="opset1">
+ <data element_type="f16" offset="16804" shape="1,16,1,1" size="32"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="590" name="bottleneck2_3/inner/dw1/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck2_3/inner/dw1/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="591" name="bottleneck2_3/inner/dw1/fn/weights3097239851904" type="Const" version="opset1">
+ <data element_type="f32" offset="1576" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="592" name="bottleneck2_3/inner/dw1/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck2_3/inner/dw1/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="593" name="bottleneck2_3/dim_inc/conv/fq_input_0" type="FakeQuantize" version="opset1">
+ <data auto_broadcast="numpy" levels="256"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1"/>
+ <port id="2"/>
+ <port id="3"/>
+ <port id="4"/>
+ </input>
+ <output>
+ <port id="5" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="594" name="bottleneck2_3/dim_inc/bn/mean/Fused_Mul__copy90610153/quantized1408019776" type="Const" version="opset1">
+ <data element_type="i8" offset="16836" shape="64,16,1,1" size="1024"/>
+ <output>
+ <port id="0" precision="I8">
+ <dim>64</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="595" name="bottleneck2_3/dim_inc/bn/mean/Fused_Mul__copy90610153/quantized/to_f16" type="Convert" version="opset1">
+ <data destination_type="f16"/>
+ <input>
+ <port id="0">
+ <dim>64</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="1" precision="FP16">
+ <dim>64</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="596" name="bottleneck2_3/dim_inc/conv/fq_weights_1/zero_point1409320313" type="Const" version="opset1">
+ <data element_type="f16" offset="17860" shape="64,1,1,1" size="128"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="597" name="bottleneck2_3/dim_inc/conv/fq_weights_1/minus_zp" type="Subtract" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>64</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="1">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>64</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="598" name="bottleneck2_3/dim_inc/conv/fq_weights_1/scale1408822134" type="Const" version="opset1">
+ <data element_type="f16" offset="17988" shape="64,1,1,1" size="128"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="599" name="bottleneck2_3/dim_inc/conv/fq_weights_1/mulpiply_by_scale" type="Multiply" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>64</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="1">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>64</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="600" name="bottleneck2_3/dim_inc/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1,1" pads_begin="0,0" pads_end="0,0" strides="1,1"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1">
+ <dim>64</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="601" name="data_add_238492385490819800" type="Const" version="opset1">
+ <data element_type="f16" offset="18116" shape="1,64,1,1" size="128"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="602" name="bottleneck2_3/dim_inc/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck2_3/dim_inc/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="603" name="bottleneck2_3/add/fq_input_1" type="FakeQuantize" version="opset1">
+ <data auto_broadcast="numpy" levels="256"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1"/>
+ <port id="2"/>
+ <port id="3"/>
+ <port id="4"/>
+ </input>
+ <output>
+ <port id="5" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="604" name="bottleneck2_3/add" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck2_3/add" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="605" name="bottleneck2_3/fn/weights3089239845911" type="Const" version="opset1">
+ <data element_type="f32" offset="1576" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="606" name="bottleneck2_3/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck2_3/add" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="607" name="bottleneck2_4/add/fq_input_0" type="FakeQuantize" version="opset1">
+ <data auto_broadcast="numpy" levels="256"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1"/>
+ <port id="2"/>
+ <port id="3"/>
+ <port id="4"/>
+ </input>
+ <output>
+ <port id="5" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="608" name="5034503822635" type="Const" version="opset1">
+ <data element_type="f16" offset="18244" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="609" name="5035503922557" type="Const" version="opset1">
+ <data element_type="f16" offset="18246" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="610" name="5036504020106" type="Const" version="opset1">
+ <data element_type="f16" offset="18244" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="611" name="5037504119368" type="Const" version="opset1">
+ <data element_type="f16" offset="18246" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="612" name="3224322820490" type="Const" version="opset1">
+ <data element_type="f16" offset="18248" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="613" name="3225322921099" type="Const" version="opset1">
+ <data element_type="f16" offset="18250" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="614" name="3226323021711" type="Const" version="opset1">
+ <data element_type="f16" offset="18248" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="615" name="3227323121966" type="Const" version="opset1">
+ <data element_type="f16" offset="18250" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="616" name="2644264820301" type="Const" version="opset1">
+ <data element_type="f16" offset="18252" shape="1,16,1,1" size="32"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="617" name="2645264919755" type="Const" version="opset1">
+ <data element_type="f16" offset="18284" shape="1,16,1,1" size="32"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="618" name="2646265021318" type="Const" version="opset1">
+ <data element_type="f16" offset="18252" shape="1,16,1,1" size="32"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="619" name="2647265121384" type="Const" version="opset1">
+ <data element_type="f16" offset="18284" shape="1,16,1,1" size="32"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="620" name="bottleneck2_4/dim_red/bn/mean/Fused_Mul__copy91310156/quantized1398421159" type="Const" version="opset1">
+ <data element_type="i8" offset="18316" shape="16,64,1,1" size="1024"/>
+ <output>
+ <port id="0" precision="I8">
+ <dim>16</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="621" name="bottleneck2_4/dim_red/bn/mean/Fused_Mul__copy91310156/quantized/to_f16" type="Convert" version="opset1">
+ <data destination_type="f16"/>
+ <input>
+ <port id="0">
+ <dim>16</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="1" precision="FP16">
+ <dim>16</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="622" name="bottleneck2_4/dim_red/conv/fq_weights_1/zero_point1399721360" type="Const" version="opset1">
+ <data element_type="f16" offset="19340" shape="16,1,1,1" size="32"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="623" name="bottleneck2_4/dim_red/conv/fq_weights_1/minus_zp" type="Subtract" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>16</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="1">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>16</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="624" name="bottleneck2_4/dim_red/conv/fq_weights_1/scale1399220787" type="Const" version="opset1">
+ <data element_type="f16" offset="19372" shape="16,1,1,1" size="32"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="625" name="bottleneck2_4/dim_red/conv/fq_weights_1/mulpiply_by_scale" type="Multiply" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>16</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="1">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>16</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="626" name="bottleneck2_4/dim_red/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1,1" pads_begin="0,0" pads_end="0,0" strides="1,1"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1">
+ <dim>16</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="627" name="data_add_238572386291522089" type="Const" version="opset1">
+ <data element_type="f16" offset="19404" shape="1,16,1,1" size="32"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="628" name="bottleneck2_4/dim_red/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck2_4/dim_red/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="629" name="bottleneck2_4/dim_red/fn/weights3114440556917" type="Const" version="opset1">
+ <data element_type="f32" offset="1576" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="630" name="bottleneck2_4/dim_red/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck2_4/dim_red/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="631" name="bottleneck2_4/inner/dw1/conv/fq_input_0" type="FakeQuantize" version="opset1">
+ <data auto_broadcast="numpy" levels="256"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="2">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="3">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="4">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="5" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="632" name="16803/value1680520367" type="Const" version="opset1">
+ <data element_type="i64" offset="8036" shape="5" size="40"/>
+ <output>
+ <port id="0" precision="I64">
+ <dim>5</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="633" name="bottleneck2_4/inner/dw1/bn/mean/Fused_Mul__copy91910159/quantized1163222761" type="Const" version="opset1">
+ <data element_type="i8" offset="19436" shape="16,1,3,3" size="144"/>
+ <output>
+ <port id="0" precision="I8">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="634" name="bottleneck2_4/inner/dw1/bn/mean/Fused_Mul__copy91910159/quantized/to_f16" type="Convert" version="opset1">
+ <data destination_type="f16"/>
+ <input>
+ <port id="0">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </input>
+ <output>
+ <port id="1" precision="FP16">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="635" name="bottleneck2_4/inner/dw1/conv/fq_weights_1/zero_point1164522191" type="Const" version="opset1">
+ <data element_type="f16" offset="19580" shape="16,1,1,1" size="32"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="636" name="bottleneck2_4/inner/dw1/conv/fq_weights_1/minus_zp" type="Subtract" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ <port id="1">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="637" name="bottleneck2_4/inner/dw1/conv/fq_weights_1/scale1164021003" type="Const" version="opset1">
+ <data element_type="f16" offset="19612" shape="16,1,1,1" size="32"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="638" name="bottleneck2_4/inner/dw1/conv/fq_weights_1/mulpiply_by_scale" type="Multiply" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ <port id="1">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="639" name="16803" type="Reshape" version="opset1">
+ <data special_zero="true"/>
+ <input>
+ <port id="0">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ <port id="1">
+ <dim>5</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="640" name="bottleneck2_4/inner/dw1/conv" type="GroupConvolution" version="opset1">
+ <data auto_pad="explicit" dilations="1,1" pads_begin="1,1" pads_end="1,1" strides="1,1"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="641" name="data_add_238652387092120916" type="Const" version="opset1">
+ <data element_type="f16" offset="19644" shape="1,16,1,1" size="32"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="642" name="bottleneck2_4/inner/dw1/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck2_4/inner/dw1/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="643" name="bottleneck2_4/inner/dw1/fn/weights3110039950923" type="Const" version="opset1">
+ <data element_type="f32" offset="1576" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="644" name="bottleneck2_4/inner/dw1/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck2_4/inner/dw1/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="645" name="bottleneck2_4/dim_inc/conv/fq_input_0" type="FakeQuantize" version="opset1">
+ <data auto_broadcast="numpy" levels="256"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1"/>
+ <port id="2"/>
+ <port id="3"/>
+ <port id="4"/>
+ </input>
+ <output>
+ <port id="5" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="646" name="bottleneck2_4/dim_inc/bn/mean/Fused_Mul__copy92510162/quantized1189622095" type="Const" version="opset1">
+ <data element_type="i8" offset="19676" shape="64,16,1,1" size="1024"/>
+ <output>
+ <port id="0" precision="I8">
+ <dim>64</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="647" name="bottleneck2_4/dim_inc/bn/mean/Fused_Mul__copy92510162/quantized/to_f16" type="Convert" version="opset1">
+ <data destination_type="f16"/>
+ <input>
+ <port id="0">
+ <dim>64</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="1" precision="FP16">
+ <dim>64</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="648" name="bottleneck2_4/dim_inc/conv/fq_weights_1/zero_point1190922224" type="Const" version="opset1">
+ <data element_type="f16" offset="20700" shape="64,1,1,1" size="128"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="649" name="bottleneck2_4/dim_inc/conv/fq_weights_1/minus_zp" type="Subtract" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>64</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="1">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>64</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="650" name="bottleneck2_4/dim_inc/conv/fq_weights_1/scale1190421480" type="Const" version="opset1">
+ <data element_type="f16" offset="20828" shape="64,1,1,1" size="128"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="651" name="bottleneck2_4/dim_inc/conv/fq_weights_1/mulpiply_by_scale" type="Multiply" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>64</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="1">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>64</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="652" name="bottleneck2_4/dim_inc/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1,1" pads_begin="0,0" pads_end="0,0" strides="1,1"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1">
+ <dim>64</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="653" name="data_add_238732387892720712" type="Const" version="opset1">
+ <data element_type="f16" offset="20956" shape="1,64,1,1" size="128"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="654" name="bottleneck2_4/dim_inc/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck2_4/dim_inc/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="655" name="bottleneck2_4/add/fq_input_1" type="FakeQuantize" version="opset1">
+ <data auto_broadcast="numpy" levels="256"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1"/>
+ <port id="2"/>
+ <port id="3"/>
+ <port id="4"/>
+ </input>
+ <output>
+ <port id="5" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="656" name="bottleneck2_4/add" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck2_4/add" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="657" name="bottleneck2_4/fn/weights3084440526930" type="Const" version="opset1">
+ <data element_type="f32" offset="1576" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="658" name="bottleneck2_4/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck2_4/add" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="659" name="bottleneck2_5/add/fq_input_0" type="FakeQuantize" version="opset1">
+ <data auto_broadcast="numpy" levels="256"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1"/>
+ <port id="2"/>
+ <port id="3"/>
+ <port id="4"/>
+ </input>
+ <output>
+ <port id="5" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="660" name="5054505821225" type="Const" version="opset1">
+ <data element_type="f16" offset="21084" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="661" name="5055505919782" type="Const" version="opset1">
+ <data element_type="f16" offset="21086" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="662" name="5056506020196" type="Const" version="opset1">
+ <data element_type="f16" offset="21084" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="663" name="5057506122386" type="Const" version="opset1">
+ <data element_type="f16" offset="21086" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="664" name="3204320822938" type="Const" version="opset1">
+ <data element_type="f16" offset="21088" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="665" name="3205320920421" type="Const" version="opset1">
+ <data element_type="f16" offset="21090" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="666" name="3206321020400" type="Const" version="opset1">
+ <data element_type="f16" offset="21088" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="667" name="3207321119365" type="Const" version="opset1">
+ <data element_type="f16" offset="21090" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="668" name="3064306821147" type="Const" version="opset1">
+ <data element_type="f16" offset="21092" shape="1,16,1,1" size="32"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="669" name="3065306920277" type="Const" version="opset1">
+ <data element_type="f16" offset="21124" shape="1,16,1,1" size="32"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="670" name="3066307020850" type="Const" version="opset1">
+ <data element_type="f16" offset="21092" shape="1,16,1,1" size="32"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="671" name="3067307120520" type="Const" version="opset1">
+ <data element_type="f16" offset="21124" shape="1,16,1,1" size="32"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="672" name="bottleneck2_5/dim_red/bn/mean/Fused_Mul__copy93210165/quantized1367219947" type="Const" version="opset1">
+ <data element_type="i8" offset="21156" shape="16,64,1,1" size="1024"/>
+ <output>
+ <port id="0" precision="I8">
+ <dim>16</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="673" name="bottleneck2_5/dim_red/bn/mean/Fused_Mul__copy93210165/quantized/to_f16" type="Convert" version="opset1">
+ <data destination_type="f16"/>
+ <input>
+ <port id="0">
+ <dim>16</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="1" precision="FP16">
+ <dim>16</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="674" name="bottleneck2_5/dim_red/conv/fq_weights_1/zero_point1368521183" type="Const" version="opset1">
+ <data element_type="f16" offset="22180" shape="16,1,1,1" size="32"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="675" name="bottleneck2_5/dim_red/conv/fq_weights_1/minus_zp" type="Subtract" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>16</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="1">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>16</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="676" name="bottleneck2_5/dim_red/conv/fq_weights_1/scale1368022131" type="Const" version="opset1">
+ <data element_type="f16" offset="22212" shape="16,1,1,1" size="32"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="677" name="bottleneck2_5/dim_red/conv/fq_weights_1/mulpiply_by_scale" type="Multiply" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>16</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="1">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>16</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="678" name="bottleneck2_5/dim_red/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1,1" pads_begin="0,0" pads_end="0,0" strides="1,1"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1">
+ <dim>16</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="679" name="data_add_238812388693420631" type="Const" version="opset1">
+ <data element_type="f16" offset="22244" shape="1,16,1,1" size="32"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="680" name="bottleneck2_5/dim_red/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck2_5/dim_red/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="681" name="bottleneck2_5/dim_red/fn/weights3114840367936" type="Const" version="opset1">
+ <data element_type="f32" offset="1576" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="682" name="bottleneck2_5/dim_red/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck2_5/dim_red/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="683" name="bottleneck2_5/inner/dw1/conv/fq_input_0" type="FakeQuantize" version="opset1">
+ <data auto_broadcast="numpy" levels="256"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="2">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="3">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="4">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="5" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="684" name="16811/value1681320502" type="Const" version="opset1">
+ <data element_type="i64" offset="8036" shape="5" size="40"/>
+ <output>
+ <port id="0" precision="I64">
+ <dim>5</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="685" name="bottleneck2_5/inner/dw1/bn/mean/Fused_Mul__copy93810168/quantized1348021606" type="Const" version="opset1">
+ <data element_type="i8" offset="22276" shape="16,1,3,3" size="144"/>
+ <output>
+ <port id="0" precision="I8">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="686" name="bottleneck2_5/inner/dw1/bn/mean/Fused_Mul__copy93810168/quantized/to_f16" type="Convert" version="opset1">
+ <data destination_type="f16"/>
+ <input>
+ <port id="0">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </input>
+ <output>
+ <port id="1" precision="FP16">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="687" name="bottleneck2_5/inner/dw1/conv/fq_weights_1/zero_point1349320136" type="Const" version="opset1">
+ <data element_type="f16" offset="22420" shape="16,1,1,1" size="32"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="688" name="bottleneck2_5/inner/dw1/conv/fq_weights_1/minus_zp" type="Subtract" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ <port id="1">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="689" name="bottleneck2_5/inner/dw1/conv/fq_weights_1/scale1348820358" type="Const" version="opset1">
+ <data element_type="f16" offset="22452" shape="16,1,1,1" size="32"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="690" name="bottleneck2_5/inner/dw1/conv/fq_weights_1/mulpiply_by_scale" type="Multiply" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ <port id="1">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="691" name="16811" type="Reshape" version="opset1">
+ <data special_zero="true"/>
+ <input>
+ <port id="0">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ <port id="1">
+ <dim>5</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="692" name="bottleneck2_5/inner/dw1/conv" type="GroupConvolution" version="opset1">
+ <data auto_pad="explicit" dilations="1,1" pads_begin="1,1" pads_end="1,1" strides="1,1"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="693" name="data_add_238892389494020859" type="Const" version="opset1">
+ <data element_type="f16" offset="22484" shape="1,16,1,1" size="32"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="694" name="bottleneck2_5/inner/dw1/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck2_5/inner/dw1/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="695" name="bottleneck2_5/inner/dw1/fn/weights3078440256942" type="Const" version="opset1">
+ <data element_type="f32" offset="1576" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="696" name="bottleneck2_5/inner/dw1/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck2_5/inner/dw1/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="697" name="bottleneck2_5/dim_inc/conv/fq_input_0" type="FakeQuantize" version="opset1">
+ <data auto_broadcast="numpy" levels="256"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1"/>
+ <port id="2"/>
+ <port id="3"/>
+ <port id="4"/>
+ </input>
+ <output>
+ <port id="5" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="698" name="bottleneck2_5/dim_inc/bn/mean/Fused_Mul__copy94410171/quantized1290419635" type="Const" version="opset1">
+ <data element_type="i8" offset="22516" shape="64,16,1,1" size="1024"/>
+ <output>
+ <port id="0" precision="I8">
+ <dim>64</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="699" name="bottleneck2_5/dim_inc/bn/mean/Fused_Mul__copy94410171/quantized/to_f16" type="Convert" version="opset1">
+ <data destination_type="f16"/>
+ <input>
+ <port id="0">
+ <dim>64</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="1" precision="FP16">
+ <dim>64</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="700" name="bottleneck2_5/dim_inc/conv/fq_weights_1/zero_point1291720784" type="Const" version="opset1">
+ <data element_type="f16" offset="23540" shape="64,1,1,1" size="128"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="701" name="bottleneck2_5/dim_inc/conv/fq_weights_1/minus_zp" type="Subtract" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>64</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="1">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>64</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="702" name="bottleneck2_5/dim_inc/conv/fq_weights_1/scale1291222455" type="Const" version="opset1">
+ <data element_type="f16" offset="23668" shape="64,1,1,1" size="128"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="703" name="bottleneck2_5/dim_inc/conv/fq_weights_1/mulpiply_by_scale" type="Multiply" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>64</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="1">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>64</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="704" name="bottleneck2_5/dim_inc/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1,1" pads_begin="0,0" pads_end="0,0" strides="1,1"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1">
+ <dim>64</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="705" name="data_add_238972390294621426" type="Const" version="opset1">
+ <data element_type="f16" offset="23796" shape="1,64,1,1" size="128"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="706" name="bottleneck2_5/dim_inc/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck2_5/dim_inc/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="707" name="bottleneck2_5/add/fq_input_1" type="FakeQuantize" version="opset1">
+ <data auto_broadcast="numpy" levels="256"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1"/>
+ <port id="2"/>
+ <port id="3"/>
+ <port id="4"/>
+ </input>
+ <output>
+ <port id="5" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="708" name="bottleneck2_5/add" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck2_5/add" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="709" name="bottleneck2_5/fn/weights3085640373949" type="Const" version="opset1">
+ <data element_type="f32" offset="1576" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="710" name="bottleneck2_5/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck2_5/add" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="711" name="bottleneck2_6/add/fq_input_0" type="FakeQuantize" version="opset1">
+ <data auto_broadcast="numpy" levels="256"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1"/>
+ <port id="2"/>
+ <port id="3"/>
+ <port id="4"/>
+ </input>
+ <output>
+ <port id="5" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="712" name="2914291819626" type="Const" version="opset1">
+ <data element_type="f16" offset="23924" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="713" name="2915291920778" type="Const" version="opset1">
+ <data element_type="f16" offset="23926" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="714" name="2916292022956" type="Const" version="opset1">
+ <data element_type="f16" offset="23924" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="715" name="2917292121741" type="Const" version="opset1">
+ <data element_type="f16" offset="23926" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="716" name="3984398821168" type="Const" version="opset1">
+ <data element_type="f16" offset="23928" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="717" name="3985398919848" type="Const" version="opset1">
+ <data element_type="f16" offset="23930" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="718" name="3986399020754" type="Const" version="opset1">
+ <data element_type="f16" offset="23928" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="719" name="3987399122671" type="Const" version="opset1">
+ <data element_type="f16" offset="23930" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="720" name="5464546820265" type="Const" version="opset1">
+ <data element_type="f16" offset="23932" shape="1,16,1,1" size="32"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="721" name="5465546919521" type="Const" version="opset1">
+ <data element_type="f16" offset="23964" shape="1,16,1,1" size="32"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="722" name="5466547021918" type="Const" version="opset1">
+ <data element_type="f16" offset="23932" shape="1,16,1,1" size="32"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="723" name="5467547121174" type="Const" version="opset1">
+ <data element_type="f16" offset="23964" shape="1,16,1,1" size="32"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="724" name="bottleneck2_6/dim_red/bn/mean/Fused_Mul__copy95110174/quantized1264019545" type="Const" version="opset1">
+ <data element_type="i8" offset="23996" shape="16,64,1,1" size="1024"/>
+ <output>
+ <port id="0" precision="I8">
+ <dim>16</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="725" name="bottleneck2_6/dim_red/bn/mean/Fused_Mul__copy95110174/quantized/to_f16" type="Convert" version="opset1">
+ <data destination_type="f16"/>
+ <input>
+ <port id="0">
+ <dim>16</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="1" precision="FP16">
+ <dim>16</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="726" name="bottleneck2_6/dim_red/conv/fq_weights_1/zero_point1265322986" type="Const" version="opset1">
+ <data element_type="f16" offset="25020" shape="16,1,1,1" size="32"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="727" name="bottleneck2_6/dim_red/conv/fq_weights_1/minus_zp" type="Subtract" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>16</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="1">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>16</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="728" name="bottleneck2_6/dim_red/conv/fq_weights_1/scale1264821084" type="Const" version="opset1">
+ <data element_type="f16" offset="25052" shape="16,1,1,1" size="32"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="729" name="bottleneck2_6/dim_red/conv/fq_weights_1/mulpiply_by_scale" type="Multiply" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>16</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="1">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>16</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="730" name="bottleneck2_6/dim_red/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1,1" pads_begin="0,0" pads_end="0,0" strides="1,1"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1">
+ <dim>16</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="731" name="data_add_239052391095321834" type="Const" version="opset1">
+ <data element_type="f16" offset="25084" shape="1,16,1,1" size="32"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="732" name="bottleneck2_6/dim_red/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck2_6/dim_red/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="733" name="bottleneck2_6/dim_red/fn/weights3083640613955" type="Const" version="opset1">
+ <data element_type="f32" offset="1576" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="734" name="bottleneck2_6/dim_red/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck2_6/dim_red/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="735" name="bottleneck2_6/inner/dw1/conv/fq_input_0" type="FakeQuantize" version="opset1">
+ <data auto_broadcast="numpy" levels="256"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="2">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="3">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="4">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="5" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="736" name="16931/value1693322713" type="Const" version="opset1">
+ <data element_type="i64" offset="8036" shape="5" size="40"/>
+ <output>
+ <port id="0" precision="I64">
+ <dim>5</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="737" name="bottleneck2_6/inner/dw1/bn/mean/Fused_Mul__copy95710177/quantized1388819623" type="Const" version="opset1">
+ <data element_type="i8" offset="25116" shape="16,1,3,3" size="144"/>
+ <output>
+ <port id="0" precision="I8">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="738" name="bottleneck2_6/inner/dw1/bn/mean/Fused_Mul__copy95710177/quantized/to_f16" type="Convert" version="opset1">
+ <data destination_type="f16"/>
+ <input>
+ <port id="0">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </input>
+ <output>
+ <port id="1" precision="FP16">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="739" name="bottleneck2_6/inner/dw1/conv/fq_weights_1/zero_point1390121903" type="Const" version="opset1">
+ <data element_type="f16" offset="25260" shape="16,1,1,1" size="32"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="740" name="bottleneck2_6/inner/dw1/conv/fq_weights_1/minus_zp" type="Subtract" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ <port id="1">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="741" name="bottleneck2_6/inner/dw1/conv/fq_weights_1/scale1389619968" type="Const" version="opset1">
+ <data element_type="f16" offset="25292" shape="16,1,1,1" size="32"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="742" name="bottleneck2_6/inner/dw1/conv/fq_weights_1/mulpiply_by_scale" type="Multiply" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ <port id="1">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="743" name="16931" type="Reshape" version="opset1">
+ <data special_zero="true"/>
+ <input>
+ <port id="0">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ <port id="1">
+ <dim>5</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="744" name="bottleneck2_6/inner/dw1/conv" type="GroupConvolution" version="opset1">
+ <data auto_pad="explicit" dilations="1,1" pads_begin="1,1" pads_end="1,1" strides="1,1"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="745" name="data_add_239132391895919890" type="Const" version="opset1">
+ <data element_type="f16" offset="25324" shape="1,16,1,1" size="32"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="746" name="bottleneck2_6/inner/dw1/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck2_6/inner/dw1/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="747" name="bottleneck2_6/inner/dw1/fn/weights3101239827961" type="Const" version="opset1">
+ <data element_type="f32" offset="1576" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="748" name="bottleneck2_6/inner/dw1/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck2_6/inner/dw1/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="749" name="bottleneck2_6/dim_inc/conv/fq_input_0" type="FakeQuantize" version="opset1">
+ <data auto_broadcast="numpy" levels="256"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1"/>
+ <port id="2"/>
+ <port id="3"/>
+ <port id="4"/>
+ </input>
+ <output>
+ <port id="5" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="750" name="bottleneck2_6/dim_inc/bn/mean/Fused_Mul__copy96310180/quantized1309619398" type="Const" version="opset1">
+ <data element_type="i8" offset="25356" shape="64,16,1,1" size="1024"/>
+ <output>
+ <port id="0" precision="I8">
+ <dim>64</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="751" name="bottleneck2_6/dim_inc/bn/mean/Fused_Mul__copy96310180/quantized/to_f16" type="Convert" version="opset1">
+ <data destination_type="f16"/>
+ <input>
+ <port id="0">
+ <dim>64</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="1" precision="FP16">
+ <dim>64</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="752" name="bottleneck2_6/dim_inc/conv/fq_weights_1/zero_point1310921357" type="Const" version="opset1">
+ <data element_type="f16" offset="26380" shape="64,1,1,1" size="128"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="753" name="bottleneck2_6/dim_inc/conv/fq_weights_1/minus_zp" type="Subtract" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>64</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="1">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>64</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="754" name="bottleneck2_6/dim_inc/conv/fq_weights_1/scale1310421198" type="Const" version="opset1">
+ <data element_type="f16" offset="26508" shape="64,1,1,1" size="128"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="755" name="bottleneck2_6/dim_inc/conv/fq_weights_1/mulpiply_by_scale" type="Multiply" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>64</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="1">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>64</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="756" name="bottleneck2_6/dim_inc/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1,1" pads_begin="0,0" pads_end="0,0" strides="1,1"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1">
+ <dim>64</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="757" name="data_add_239212392696522050" type="Const" version="opset1">
+ <data element_type="f16" offset="26636" shape="1,64,1,1" size="128"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="758" name="bottleneck2_6/dim_inc/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck2_6/dim_inc/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="759" name="bottleneck2_6/add/fq_input_1" type="FakeQuantize" version="opset1">
+ <data auto_broadcast="numpy" levels="256"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1"/>
+ <port id="2"/>
+ <port id="3"/>
+ <port id="4"/>
+ </input>
+ <output>
+ <port id="5" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="760" name="bottleneck2_6/add" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck2_6/add" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="761" name="bottleneck2_6/fn/weights3076040430968" type="Const" version="opset1">
+ <data element_type="f32" offset="1576" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="762" name="bottleneck2_6/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck2_6/add" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="763" name="bottleneck2_7/add/fq_input_0" type="FakeQuantize" version="opset1">
+ <data auto_broadcast="numpy" levels="256"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1"/>
+ <port id="2"/>
+ <port id="3"/>
+ <port id="4"/>
+ </input>
+ <output>
+ <port id="5" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="764" name="4454445822257" type="Const" version="opset1">
+ <data element_type="f16" offset="26764" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="765" name="4455445921093" type="Const" version="opset1">
+ <data element_type="f16" offset="26766" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="766" name="4456446019599" type="Const" version="opset1">
+ <data element_type="f16" offset="26764" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="767" name="4457446122842" type="Const" version="opset1">
+ <data element_type="f16" offset="26766" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="768" name="5364536821585" type="Const" version="opset1">
+ <data element_type="f16" offset="26768" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="769" name="5365536921744" type="Const" version="opset1">
+ <data element_type="f16" offset="26770" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="770" name="5366537020397" type="Const" version="opset1">
+ <data element_type="f16" offset="26768" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="771" name="5367537121087" type="Const" version="opset1">
+ <data element_type="f16" offset="26770" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="772" name="4884488819671" type="Const" version="opset1">
+ <data element_type="f16" offset="26772" shape="1,16,1,1" size="32"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="773" name="4885488919494" type="Const" version="opset1">
+ <data element_type="f16" offset="26804" shape="1,16,1,1" size="32"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="774" name="4886489022326" type="Const" version="opset1">
+ <data element_type="f16" offset="26772" shape="1,16,1,1" size="32"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="775" name="4887489119860" type="Const" version="opset1">
+ <data element_type="f16" offset="26804" shape="1,16,1,1" size="32"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="776" name="bottleneck2_7/dim_red/bn/mean/Fused_Mul__copy97010183/quantized1153619740" type="Const" version="opset1">
+ <data element_type="i8" offset="26836" shape="16,64,1,1" size="1024"/>
+ <output>
+ <port id="0" precision="I8">
+ <dim>16</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="777" name="bottleneck2_7/dim_red/bn/mean/Fused_Mul__copy97010183/quantized/to_f16" type="Convert" version="opset1">
+ <data destination_type="f16"/>
+ <input>
+ <port id="0">
+ <dim>16</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="1" precision="FP16">
+ <dim>16</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="778" name="bottleneck2_7/dim_red/conv/fq_weights_1/zero_point1154922197" type="Const" version="opset1">
+ <data element_type="f16" offset="27860" shape="16,1,1,1" size="32"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="779" name="bottleneck2_7/dim_red/conv/fq_weights_1/minus_zp" type="Subtract" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>16</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="1">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>16</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="780" name="bottleneck2_7/dim_red/conv/fq_weights_1/scale1154422755" type="Const" version="opset1">
+ <data element_type="f16" offset="27892" shape="16,1,1,1" size="32"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="781" name="bottleneck2_7/dim_red/conv/fq_weights_1/mulpiply_by_scale" type="Multiply" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>16</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="1">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>16</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="782" name="bottleneck2_7/dim_red/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1,1" pads_begin="0,0" pads_end="0,0" strides="1,1"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1">
+ <dim>16</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="783" name="data_add_239292393497221297" type="Const" version="opset1">
+ <data element_type="f16" offset="27924" shape="1,16,1,1" size="32"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="784" name="bottleneck2_7/dim_red/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck2_7/dim_red/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="785" name="bottleneck2_7/dim_red/fn/weights3110840142974" type="Const" version="opset1">
+ <data element_type="f32" offset="1576" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="786" name="bottleneck2_7/dim_red/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck2_7/dim_red/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="787" name="bottleneck2_7/inner/dw1/conv/fq_input_0" type="FakeQuantize" version="opset1">
+ <data auto_broadcast="numpy" levels="256"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="2">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="3">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="4">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="5" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="788" name="16911/value1691319764" type="Const" version="opset1">
+ <data element_type="i64" offset="8036" shape="5" size="40"/>
+ <output>
+ <port id="0" precision="I64">
+ <dim>5</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="789" name="bottleneck2_7/inner/dw1/bn/mean/Fused_Mul__copy97610186/quantized1321619503" type="Const" version="opset1">
+ <data element_type="i8" offset="27956" shape="16,1,3,3" size="144"/>
+ <output>
+ <port id="0" precision="I8">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="790" name="bottleneck2_7/inner/dw1/bn/mean/Fused_Mul__copy97610186/quantized/to_f16" type="Convert" version="opset1">
+ <data destination_type="f16"/>
+ <input>
+ <port id="0">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </input>
+ <output>
+ <port id="1" precision="FP16">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="791" name="bottleneck2_7/inner/dw1/conv/fq_weights_1/zero_point1322920703" type="Const" version="opset1">
+ <data element_type="f16" offset="28100" shape="16,1,1,1" size="32"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="792" name="bottleneck2_7/inner/dw1/conv/fq_weights_1/minus_zp" type="Subtract" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ <port id="1">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="793" name="bottleneck2_7/inner/dw1/conv/fq_weights_1/scale1322419530" type="Const" version="opset1">
+ <data element_type="f16" offset="28132" shape="16,1,1,1" size="32"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="794" name="bottleneck2_7/inner/dw1/conv/fq_weights_1/mulpiply_by_scale" type="Multiply" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ <port id="1">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="795" name="16911" type="Reshape" version="opset1">
+ <data special_zero="true"/>
+ <input>
+ <port id="0">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ <port id="1">
+ <dim>5</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="796" name="bottleneck2_7/inner/dw1/conv" type="GroupConvolution" version="opset1">
+ <data auto_pad="explicit" dilations="1,1" pads_begin="1,1" pads_end="1,1" strides="1,1"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="797" name="data_add_239372394297821825" type="Const" version="opset1">
+ <data element_type="f16" offset="28164" shape="1,16,1,1" size="32"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="798" name="bottleneck2_7/inner/dw1/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck2_7/inner/dw1/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="799" name="bottleneck2_7/inner/dw1/fn/weights3088040439980" type="Const" version="opset1">
+ <data element_type="f32" offset="1576" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="800" name="bottleneck2_7/inner/dw1/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck2_7/inner/dw1/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="801" name="bottleneck2_7/dim_inc/conv/fq_input_0" type="FakeQuantize" version="opset1">
+ <data auto_broadcast="numpy" levels="256"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1"/>
+ <port id="2"/>
+ <port id="3"/>
+ <port id="4"/>
+ </input>
+ <output>
+ <port id="5" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="802" name="bottleneck2_7/dim_inc/bn/mean/Fused_Mul__copy98210189/quantized1211221822" type="Const" version="opset1">
+ <data element_type="i8" offset="28196" shape="64,16,1,1" size="1024"/>
+ <output>
+ <port id="0" precision="I8">
+ <dim>64</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="803" name="bottleneck2_7/dim_inc/bn/mean/Fused_Mul__copy98210189/quantized/to_f16" type="Convert" version="opset1">
+ <data destination_type="f16"/>
+ <input>
+ <port id="0">
+ <dim>64</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="1" precision="FP16">
+ <dim>64</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="804" name="bottleneck2_7/dim_inc/conv/fq_weights_1/zero_point1212522542" type="Const" version="opset1">
+ <data element_type="f16" offset="29220" shape="64,1,1,1" size="128"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="805" name="bottleneck2_7/dim_inc/conv/fq_weights_1/minus_zp" type="Subtract" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>64</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="1">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>64</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="806" name="bottleneck2_7/dim_inc/conv/fq_weights_1/scale1212022902" type="Const" version="opset1">
+ <data element_type="f16" offset="29348" shape="64,1,1,1" size="128"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="807" name="bottleneck2_7/dim_inc/conv/fq_weights_1/mulpiply_by_scale" type="Multiply" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>64</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="1">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>64</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="808" name="bottleneck2_7/dim_inc/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1,1" pads_begin="0,0" pads_end="0,0" strides="1,1"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1">
+ <dim>64</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="809" name="data_add_239452395098422182" type="Const" version="opset1">
+ <data element_type="f16" offset="29476" shape="1,64,1,1" size="128"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="810" name="bottleneck2_7/dim_inc/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck2_7/dim_inc/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="811" name="bottleneck2_7/add/fq_input_1" type="FakeQuantize" version="opset1">
+ <data auto_broadcast="numpy" levels="256"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1"/>
+ <port id="2"/>
+ <port id="3"/>
+ <port id="4"/>
+ </input>
+ <output>
+ <port id="5" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="812" name="bottleneck2_7/add" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck2_7/add" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="813" name="bottleneck2_7/fn/weights3091640094987" type="Const" version="opset1">
+ <data element_type="f32" offset="1576" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="814" name="bottleneck2_7/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck2_7/add" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="815" name="bottleneck2_8/add/fq_input_0" type="FakeQuantize" version="opset1">
+ <data auto_broadcast="numpy" levels="256"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1"/>
+ <port id="2"/>
+ <port id="3"/>
+ <port id="4"/>
+ </input>
+ <output>
+ <port id="5" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="816" name="2754275821450" type="Const" version="opset1">
+ <data element_type="f16" offset="29604" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="817" name="2755275920847" type="Const" version="opset1">
+ <data element_type="f16" offset="29606" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="818" name="2756276022839" type="Const" version="opset1">
+ <data element_type="f16" offset="29604" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="819" name="2757276121993" type="Const" version="opset1">
+ <data element_type="f16" offset="29606" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="820" name="2544254820079" type="Const" version="opset1">
+ <data element_type="f16" offset="29608" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="821" name="2545254920676" type="Const" version="opset1">
+ <data element_type="f16" offset="29610" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="822" name="2546255021231" type="Const" version="opset1">
+ <data element_type="f16" offset="29608" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="823" name="2547255121672" type="Const" version="opset1">
+ <data element_type="f16" offset="29610" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="824" name="3904390821567" type="Const" version="opset1">
+ <data element_type="f16" offset="29612" shape="1,16,1,1" size="32"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="825" name="3905390921915" type="Const" version="opset1">
+ <data element_type="f16" offset="29644" shape="1,16,1,1" size="32"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="826" name="3906391019995" type="Const" version="opset1">
+ <data element_type="f16" offset="29612" shape="1,16,1,1" size="32"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="827" name="3907391120883" type="Const" version="opset1">
+ <data element_type="f16" offset="29644" shape="1,16,1,1" size="32"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="828" name="bottleneck2_8/dim_red/bn/mean/Fused_Mul__copy98910192/quantized1316821282" type="Const" version="opset1">
+ <data element_type="i8" offset="29676" shape="16,64,1,1" size="1024"/>
+ <output>
+ <port id="0" precision="I8">
+ <dim>16</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="829" name="bottleneck2_8/dim_red/bn/mean/Fused_Mul__copy98910192/quantized/to_f16" type="Convert" version="opset1">
+ <data destination_type="f16"/>
+ <input>
+ <port id="0">
+ <dim>16</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="1" precision="FP16">
+ <dim>16</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="830" name="bottleneck2_8/dim_red/conv/fq_weights_1/zero_point1318122794" type="Const" version="opset1">
+ <data element_type="f16" offset="30700" shape="16,1,1,1" size="32"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="831" name="bottleneck2_8/dim_red/conv/fq_weights_1/minus_zp" type="Subtract" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>16</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="1">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>16</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="832" name="bottleneck2_8/dim_red/conv/fq_weights_1/scale1317620826" type="Const" version="opset1">
+ <data element_type="f16" offset="30732" shape="16,1,1,1" size="32"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="833" name="bottleneck2_8/dim_red/conv/fq_weights_1/mulpiply_by_scale" type="Multiply" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>16</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="1">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>16</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="834" name="bottleneck2_8/dim_red/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1,1" pads_begin="0,0" pads_end="0,0" strides="1,1"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1">
+ <dim>16</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="835" name="data_add_239532395899121927" type="Const" version="opset1">
+ <data element_type="f16" offset="30764" shape="1,16,1,1" size="32"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="836" name="bottleneck2_8/dim_red/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck2_8/dim_red/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="837" name="bottleneck2_8/dim_red/fn/weights3095240220993" type="Const" version="opset1">
+ <data element_type="f32" offset="1576" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="838" name="bottleneck2_8/dim_red/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck2_8/dim_red/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="839" name="bottleneck2_8/inner/dw1/conv/fq_input_0" type="FakeQuantize" version="opset1">
+ <data auto_broadcast="numpy" levels="256"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="2">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="3">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="4">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="5" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="840" name="16863/value1686522035" type="Const" version="opset1">
+ <data element_type="i64" offset="8036" shape="5" size="40"/>
+ <output>
+ <port id="0" precision="I64">
+ <dim>5</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="841" name="bottleneck2_8/inner/dw1/bn/mean/Fused_Mul__copy99510195/quantized1247221654" type="Const" version="opset1">
+ <data element_type="i8" offset="30796" shape="16,1,3,3" size="144"/>
+ <output>
+ <port id="0" precision="I8">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="842" name="bottleneck2_8/inner/dw1/bn/mean/Fused_Mul__copy99510195/quantized/to_f16" type="Convert" version="opset1">
+ <data destination_type="f16"/>
+ <input>
+ <port id="0">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </input>
+ <output>
+ <port id="1" precision="FP16">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="843" name="bottleneck2_8/inner/dw1/conv/fq_weights_1/zero_point1248519572" type="Const" version="opset1">
+ <data element_type="f16" offset="30940" shape="16,1,1,1" size="32"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="844" name="bottleneck2_8/inner/dw1/conv/fq_weights_1/minus_zp" type="Subtract" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ <port id="1">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="845" name="bottleneck2_8/inner/dw1/conv/fq_weights_1/scale1248022083" type="Const" version="opset1">
+ <data element_type="f16" offset="30972" shape="16,1,1,1" size="32"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="846" name="bottleneck2_8/inner/dw1/conv/fq_weights_1/mulpiply_by_scale" type="Multiply" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ <port id="1">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="847" name="16863" type="Reshape" version="opset1">
+ <data special_zero="true"/>
+ <input>
+ <port id="0">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ <port id="1">
+ <dim>5</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="848" name="bottleneck2_8/inner/dw1/conv" type="GroupConvolution" version="opset1">
+ <data auto_pad="explicit" dilations="1,1" pads_begin="1,1" pads_end="1,1" strides="1,1"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="849" name="data_add_239612396699719605" type="Const" version="opset1">
+ <data element_type="f16" offset="31004" shape="1,16,1,1" size="32"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="850" name="bottleneck2_8/inner/dw1/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck2_8/inner/dw1/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="851" name="bottleneck2_8/inner/dw1/fn/weights3080040325999" type="Const" version="opset1">
+ <data element_type="f32" offset="1576" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="852" name="bottleneck2_8/inner/dw1/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck2_8/inner/dw1/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="853" name="bottleneck2_8/dim_inc/conv/fq_input_0" type="FakeQuantize" version="opset1">
+ <data auto_broadcast="numpy" levels="256"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1"/>
+ <port id="2"/>
+ <port id="3"/>
+ <port id="4"/>
+ </input>
+ <output>
+ <port id="5" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="854" name="bottleneck2_8/dim_inc/bn/mean/Fused_Mul__copy100110198/quantized1355222461" type="Const" version="opset1">
+ <data element_type="i8" offset="31036" shape="64,16,1,1" size="1024"/>
+ <output>
+ <port id="0" precision="I8">
+ <dim>64</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="855" name="bottleneck2_8/dim_inc/bn/mean/Fused_Mul__copy100110198/quantized/to_f16" type="Convert" version="opset1">
+ <data destination_type="f16"/>
+ <input>
+ <port id="0">
+ <dim>64</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="1" precision="FP16">
+ <dim>64</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="856" name="bottleneck2_8/dim_inc/conv/fq_weights_1/zero_point1356520682" type="Const" version="opset1">
+ <data element_type="f16" offset="32060" shape="64,1,1,1" size="128"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="857" name="bottleneck2_8/dim_inc/conv/fq_weights_1/minus_zp" type="Subtract" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>64</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="1">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>64</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="858" name="bottleneck2_8/dim_inc/conv/fq_weights_1/scale1356022122" type="Const" version="opset1">
+ <data element_type="f16" offset="32188" shape="64,1,1,1" size="128"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="859" name="bottleneck2_8/dim_inc/conv/fq_weights_1/mulpiply_by_scale" type="Multiply" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>64</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="1">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>64</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="860" name="bottleneck2_8/dim_inc/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1,1" pads_begin="0,0" pads_end="0,0" strides="1,1"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1">
+ <dim>64</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="861" name="data_add_2396923974100320613" type="Const" version="opset1">
+ <data element_type="f16" offset="32316" shape="1,64,1,1" size="128"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="862" name="bottleneck2_8/dim_inc/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck2_8/dim_inc/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="863" name="bottleneck2_8/add/fq_input_1" type="FakeQuantize" version="opset1">
+ <data auto_broadcast="numpy" levels="256"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1"/>
+ <port id="2"/>
+ <port id="3"/>
+ <port id="4"/>
+ </input>
+ <output>
+ <port id="5" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="864" name="bottleneck2_8/add" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck2_8/add" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="865" name="bottleneck2_8/fn/weights30840401601006" type="Const" version="opset1">
+ <data element_type="f32" offset="1576" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="866" name="bottleneck2_8/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck2_8/add" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="867" name="bottleneck3_0/dim_red/conv/fq_input_0" type="FakeQuantize" version="opset1">
+ <data auto_broadcast="numpy" levels="256"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1"/>
+ <port id="2"/>
+ <port id="3"/>
+ <port id="4"/>
+ </input>
+ <output>
+ <port id="5" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="868" name="bottleneck3_0/skip/pooling" type="MaxPool" version="opset1">
+ <data auto_pad="explicit" kernel="2,2" pads_begin="0,0" pads_end="0,0" rounding_type="ceil" strides="2,2"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </input>
+ <output>
+ <port id="1" names="bottleneck3_0/skip/pooling" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="869" name="bottleneck3_0/skip/bn/mean/Fused_Mul__copy100910201/quantized1268822494" type="Const" version="opset1">
+ <data element_type="i8" offset="32444" shape="128,64,1,1" size="8192"/>
+ <output>
+ <port id="0" precision="I8">
+ <dim>128</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="870" name="bottleneck3_0/skip/bn/mean/Fused_Mul__copy100910201/quantized/to_f16" type="Convert" version="opset1">
+ <data destination_type="f16"/>
+ <input>
+ <port id="0">
+ <dim>128</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="1" precision="FP16">
+ <dim>128</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="871" name="bottleneck3_0/skip/conv/fq_weights_1/zero_point1270122716" type="Const" version="opset1">
+ <data element_type="f16" offset="40636" shape="128,1,1,1" size="256"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="872" name="bottleneck3_0/skip/conv/fq_weights_1/minus_zp" type="Subtract" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>128</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="1">
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>128</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="873" name="bottleneck3_0/skip/conv/fq_weights_1/scale1269620070" type="Const" version="opset1">
+ <data element_type="f16" offset="40892" shape="128,1,1,1" size="256"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="874" name="bottleneck3_0/skip/conv/fq_weights_1/mulpiply_by_scale" type="Multiply" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>128</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="1">
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>128</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="875" name="bottleneck3_0/skip/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1,1" pads_begin="0,0" pads_end="0,0" strides="1,1"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1">
+ <dim>128</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="876" name="data_add_2397723982101122173" type="Const" version="opset1">
+ <data element_type="f16" offset="41148" shape="1,128,1,1" size="256"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="877" name="bottleneck3_0/skip/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_0/skip/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="878" name="bottleneck3_0/add/fq_input_0" type="FakeQuantize" version="opset1">
+ <data auto_broadcast="numpy" levels="256"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1"/>
+ <port id="2"/>
+ <port id="3"/>
+ <port id="4"/>
+ </input>
+ <output>
+ <port id="5" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="879" name="4954495819584" type="Const" version="opset1">
+ <data element_type="f16" offset="41404" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="880" name="4955495920154" type="Const" version="opset1">
+ <data element_type="f16" offset="41406" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="881" name="4956496021789" type="Const" version="opset1">
+ <data element_type="f16" offset="41404" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="882" name="4957496120331" type="Const" version="opset1">
+ <data element_type="f16" offset="41406" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="883" name="3784378820817" type="Const" version="opset1">
+ <data element_type="f16" offset="41408" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="884" name="3785378922965" type="Const" version="opset1">
+ <data element_type="f16" offset="23930" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="885" name="3786379022008" type="Const" version="opset1">
+ <data element_type="f16" offset="41408" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="886" name="3787379119818" type="Const" version="opset1">
+ <data element_type="f16" offset="23930" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="887" name="3704370822284" type="Const" version="opset1">
+ <data element_type="f16" offset="41410" shape="1,32,1,1" size="64"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="888" name="3705370921024" type="Const" version="opset1">
+ <data element_type="f16" offset="41474" shape="1,32,1,1" size="64"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="889" name="3706371019812" type="Const" version="opset1">
+ <data element_type="f16" offset="41410" shape="1,32,1,1" size="64"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="890" name="3707371120835" type="Const" version="opset1">
+ <data element_type="f16" offset="41474" shape="1,32,1,1" size="64"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="891" name="bottleneck3_0/dim_red/bn/mean/Fused_Mul__copy101310203/quantized1237621900" type="Const" version="opset1">
+ <data element_type="i8" offset="41538" shape="32,64,1,1" size="2048"/>
+ <output>
+ <port id="0" precision="I8">
+ <dim>32</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="892" name="bottleneck3_0/dim_red/bn/mean/Fused_Mul__copy101310203/quantized/to_f16" type="Convert" version="opset1">
+ <data destination_type="f16"/>
+ <input>
+ <port id="0">
+ <dim>32</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="1" precision="FP16">
+ <dim>32</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="893" name="bottleneck3_0/dim_red/conv/fq_weights_1/zero_point1238919839" type="Const" version="opset1">
+ <data element_type="f16" offset="43586" shape="32,1,1,1" size="64"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="894" name="bottleneck3_0/dim_red/conv/fq_weights_1/minus_zp" type="Subtract" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>32</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="1">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>32</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="895" name="bottleneck3_0/dim_red/conv/fq_weights_1/scale1238422143" type="Const" version="opset1">
+ <data element_type="f16" offset="43650" shape="32,1,1,1" size="64"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="896" name="bottleneck3_0/dim_red/conv/fq_weights_1/mulpiply_by_scale" type="Multiply" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>32</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="1">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>32</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="897" name="bottleneck3_0/dim_red/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1,1" pads_begin="0,0" pads_end="0,0" strides="1,1"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1">
+ <dim>32</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="898" name="data_add_2398523990101520202" type="Const" version="opset1">
+ <data element_type="f16" offset="43714" shape="1,32,1,1" size="64"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="899" name="bottleneck3_0/dim_red/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_0/dim_red/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="900" name="bottleneck3_0/dim_red/fn/weights30808397941017" type="Const" version="opset1">
+ <data element_type="f32" offset="1576" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="901" name="bottleneck3_0/dim_red/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_0/dim_red/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="902" name="bottleneck3_0/inner/dw1/conv/fq_input_0" type="FakeQuantize" version="opset1">
+ <data auto_broadcast="numpy" levels="256"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="2">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="3">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="4">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="5" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="903" name="16847/value1684921366" type="Const" version="opset1">
+ <data element_type="i64" offset="43778" shape="5" size="40"/>
+ <output>
+ <port id="0" precision="I64">
+ <dim>5</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="904" name="bottleneck3_0/inner/dw1/bn/mean/Fused_Mul__copy101910206/quantized1297622350" type="Const" version="opset1">
+ <data element_type="i8" offset="43818" shape="32,1,3,3" size="288"/>
+ <output>
+ <port id="0" precision="I8">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="905" name="bottleneck3_0/inner/dw1/bn/mean/Fused_Mul__copy101910206/quantized/to_f16" type="Convert" version="opset1">
+ <data destination_type="f16"/>
+ <input>
+ <port id="0">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </input>
+ <output>
+ <port id="1" precision="FP16">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="906" name="bottleneck3_0/inner/dw1/conv/fq_weights_1/zero_point1298919377" type="Const" version="opset1">
+ <data element_type="f16" offset="44106" shape="32,1,1,1" size="64"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="907" name="bottleneck3_0/inner/dw1/conv/fq_weights_1/minus_zp" type="Subtract" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ <port id="1">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="908" name="bottleneck3_0/inner/dw1/conv/fq_weights_1/scale1298421720" type="Const" version="opset1">
+ <data element_type="f16" offset="44170" shape="32,1,1,1" size="64"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="909" name="bottleneck3_0/inner/dw1/conv/fq_weights_1/mulpiply_by_scale" type="Multiply" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ <port id="1">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="910" name="16847" type="Reshape" version="opset1">
+ <data special_zero="true"/>
+ <input>
+ <port id="0">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ <port id="1">
+ <dim>5</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="911" name="bottleneck3_0/inner/dw1/conv" type="GroupConvolution" version="opset1">
+ <data auto_pad="explicit" dilations="1,1" pads_begin="1,1" pads_end="1,1" strides="2,2"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="912" name="data_add_2399323998102121312" type="Const" version="opset1">
+ <data element_type="f16" offset="44234" shape="1,32,1,1" size="64"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="913" name="bottleneck3_0/inner/dw1/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_0/inner/dw1/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="914" name="bottleneck3_0/inner/dw1/fn/weights31172399591023" type="Const" version="opset1">
+ <data element_type="f32" offset="1576" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="915" name="bottleneck3_0/inner/dw1/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_0/inner/dw1/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="916" name="bottleneck3_0/dim_inc/conv/fq_input_0" type="FakeQuantize" version="opset1">
+ <data auto_broadcast="numpy" levels="256"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1"/>
+ <port id="2"/>
+ <port id="3"/>
+ <port id="4"/>
+ </input>
+ <output>
+ <port id="5" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="917" name="bottleneck3_0/dim_inc/bn/mean/Fused_Mul__copy102510209/quantized1288019926" type="Const" version="opset1">
+ <data element_type="i8" offset="44298" shape="128,32,1,1" size="4096"/>
+ <output>
+ <port id="0" precision="I8">
+ <dim>128</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="918" name="bottleneck3_0/dim_inc/bn/mean/Fused_Mul__copy102510209/quantized/to_f16" type="Convert" version="opset1">
+ <data destination_type="f16"/>
+ <input>
+ <port id="0">
+ <dim>128</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="1" precision="FP16">
+ <dim>128</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="919" name="bottleneck3_0/dim_inc/conv/fq_weights_1/zero_point1289320658" type="Const" version="opset1">
+ <data element_type="f16" offset="48394" shape="128,1,1,1" size="256"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="920" name="bottleneck3_0/dim_inc/conv/fq_weights_1/minus_zp" type="Subtract" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>128</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="1">
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>128</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="921" name="bottleneck3_0/dim_inc/conv/fq_weights_1/scale1288819863" type="Const" version="opset1">
+ <data element_type="f16" offset="48650" shape="128,1,1,1" size="256"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="922" name="bottleneck3_0/dim_inc/conv/fq_weights_1/mulpiply_by_scale" type="Multiply" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>128</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="1">
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>128</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="923" name="bottleneck3_0/dim_inc/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1,1" pads_begin="0,0" pads_end="0,0" strides="1,1"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1">
+ <dim>128</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="924" name="data_add_2400124006102719554" type="Const" version="opset1">
+ <data element_type="f16" offset="48906" shape="1,128,1,1" size="256"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="925" name="bottleneck3_0/dim_inc/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_0/dim_inc/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="926" name="bottleneck3_0/add/fq_input_1" type="FakeQuantize" version="opset1">
+ <data auto_broadcast="numpy" levels="256"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1"/>
+ <port id="2"/>
+ <port id="3"/>
+ <port id="4"/>
+ </input>
+ <output>
+ <port id="5" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="927" name="bottleneck3_0/add" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_0/add" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="928" name="bottleneck3_0/fn/weights31120402591030" type="Const" version="opset1">
+ <data element_type="f32" offset="1576" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="929" name="bottleneck3_0/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_0/add" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="930" name="bottleneck3_1/add/fq_input_0" type="FakeQuantize" version="opset1">
+ <data auto_broadcast="numpy" levels="256"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1"/>
+ <port id="2"/>
+ <port id="3"/>
+ <port id="4"/>
+ </input>
+ <output>
+ <port id="5" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="931" name="5214521820256" type="Const" version="opset1">
+ <data element_type="f16" offset="49162" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="932" name="5215521920289" type="Const" version="opset1">
+ <data element_type="f16" offset="49164" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="933" name="5216522020670" type="Const" version="opset1">
+ <data element_type="f16" offset="49162" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="934" name="5217522120055" type="Const" version="opset1">
+ <data element_type="f16" offset="49164" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="935" name="2604260820451" type="Const" version="opset1">
+ <data element_type="f16" offset="49166" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="936" name="2605260922074" type="Const" version="opset1">
+ <data element_type="f16" offset="49168" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="937" name="2606261022275" type="Const" version="opset1">
+ <data element_type="f16" offset="49166" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="938" name="2607261120814" type="Const" version="opset1">
+ <data element_type="f16" offset="49168" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="939" name="3684368822539" type="Const" version="opset1">
+ <data element_type="f16" offset="49170" shape="1,32,1,1" size="64"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="940" name="3685368922827" type="Const" version="opset1">
+ <data element_type="f16" offset="49234" shape="1,32,1,1" size="64"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="941" name="3686369020463" type="Const" version="opset1">
+ <data element_type="f16" offset="49170" shape="1,32,1,1" size="64"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="942" name="3687369120550" type="Const" version="opset1">
+ <data element_type="f16" offset="49234" shape="1,32,1,1" size="64"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="943" name="bottleneck3_1/dim_red/bn/mean/Fused_Mul__copy103210212/quantized1400819425" type="Const" version="opset1">
+ <data element_type="i8" offset="49298" shape="32,128,1,1" size="4096"/>
+ <output>
+ <port id="0" precision="I8">
+ <dim>32</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="944" name="bottleneck3_1/dim_red/bn/mean/Fused_Mul__copy103210212/quantized/to_f16" type="Convert" version="opset1">
+ <data destination_type="f16"/>
+ <input>
+ <port id="0">
+ <dim>32</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="1" precision="FP16">
+ <dim>32</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="945" name="bottleneck3_1/dim_red/conv/fq_weights_1/zero_point1402122161" type="Const" version="opset1">
+ <data element_type="f16" offset="53394" shape="32,1,1,1" size="64"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="946" name="bottleneck3_1/dim_red/conv/fq_weights_1/minus_zp" type="Subtract" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>32</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="1">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>32</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="947" name="bottleneck3_1/dim_red/conv/fq_weights_1/scale1401622887" type="Const" version="opset1">
+ <data element_type="f16" offset="53458" shape="32,1,1,1" size="64"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="948" name="bottleneck3_1/dim_red/conv/fq_weights_1/mulpiply_by_scale" type="Multiply" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>32</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="1">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>32</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="949" name="bottleneck3_1/dim_red/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1,1" pads_begin="0,0" pads_end="0,0" strides="1,1"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1">
+ <dim>32</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="950" name="data_add_2400924014103419488" type="Const" version="opset1">
+ <data element_type="f16" offset="53522" shape="1,32,1,1" size="64"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="951" name="bottleneck3_1/dim_red/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_1/dim_red/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="952" name="bottleneck3_1/dim_red/fn/weights31008405681036" type="Const" version="opset1">
+ <data element_type="f32" offset="1576" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="953" name="bottleneck3_1/dim_red/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_1/dim_red/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="954" name="bottleneck3_1/inner/dw1/conv/fq_input_0" type="FakeQuantize" version="opset1">
+ <data auto_broadcast="numpy" levels="256"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="2">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="3">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="4">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="5" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="955" name="16843/value1684521144" type="Const" version="opset1">
+ <data element_type="i64" offset="43778" shape="5" size="40"/>
+ <output>
+ <port id="0" precision="I64">
+ <dim>5</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="956" name="bottleneck3_1/inner/dw1/bn/mean/Fused_Mul__copy103810215/quantized1283219938" type="Const" version="opset1">
+ <data element_type="i8" offset="53586" shape="32,1,3,3" size="288"/>
+ <output>
+ <port id="0" precision="I8">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="957" name="bottleneck3_1/inner/dw1/bn/mean/Fused_Mul__copy103810215/quantized/to_f16" type="Convert" version="opset1">
+ <data destination_type="f16"/>
+ <input>
+ <port id="0">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </input>
+ <output>
+ <port id="1" precision="FP16">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="958" name="bottleneck3_1/inner/dw1/conv/fq_weights_1/zero_point1284522812" type="Const" version="opset1">
+ <data element_type="f16" offset="53874" shape="32,1,1,1" size="64"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="959" name="bottleneck3_1/inner/dw1/conv/fq_weights_1/minus_zp" type="Subtract" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ <port id="1">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="960" name="bottleneck3_1/inner/dw1/conv/fq_weights_1/scale1284019992" type="Const" version="opset1">
+ <data element_type="f16" offset="53938" shape="32,1,1,1" size="64"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="961" name="bottleneck3_1/inner/dw1/conv/fq_weights_1/mulpiply_by_scale" type="Multiply" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ <port id="1">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="962" name="16843" type="Reshape" version="opset1">
+ <data special_zero="true"/>
+ <input>
+ <port id="0">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ <port id="1">
+ <dim>5</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="963" name="bottleneck3_1/inner/dw1/conv" type="GroupConvolution" version="opset1">
+ <data auto_pad="explicit" dilations="1,1" pads_begin="1,1" pads_end="1,1" strides="1,1"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="964" name="data_add_2401724022104019401" type="Const" version="opset1">
+ <data element_type="f16" offset="54002" shape="1,32,1,1" size="64"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="965" name="bottleneck3_1/inner/dw1/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_1/inner/dw1/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="966" name="bottleneck3_1/inner/dw1/fn/weights30968403911042" type="Const" version="opset1">
+ <data element_type="f32" offset="1576" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="967" name="bottleneck3_1/inner/dw1/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_1/inner/dw1/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="968" name="bottleneck3_1/dim_inc/conv/fq_input_0" type="FakeQuantize" version="opset1">
+ <data auto_broadcast="numpy" levels="256"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1"/>
+ <port id="2"/>
+ <port id="3"/>
+ <port id="4"/>
+ </input>
+ <output>
+ <port id="5" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="969" name="bottleneck3_1/dim_inc/bn/mean/Fused_Mul__copy104410218/quantized1307222638" type="Const" version="opset1">
+ <data element_type="i8" offset="54066" shape="128,32,1,1" size="4096"/>
+ <output>
+ <port id="0" precision="I8">
+ <dim>128</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="970" name="bottleneck3_1/dim_inc/bn/mean/Fused_Mul__copy104410218/quantized/to_f16" type="Convert" version="opset1">
+ <data destination_type="f16"/>
+ <input>
+ <port id="0">
+ <dim>128</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="1" precision="FP16">
+ <dim>128</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="971" name="bottleneck3_1/dim_inc/conv/fq_weights_1/zero_point1308520343" type="Const" version="opset1">
+ <data element_type="f16" offset="58162" shape="128,1,1,1" size="256"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="972" name="bottleneck3_1/dim_inc/conv/fq_weights_1/minus_zp" type="Subtract" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>128</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="1">
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>128</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="973" name="bottleneck3_1/dim_inc/conv/fq_weights_1/scale1308019578" type="Const" version="opset1">
+ <data element_type="f16" offset="58418" shape="128,1,1,1" size="256"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="974" name="bottleneck3_1/dim_inc/conv/fq_weights_1/mulpiply_by_scale" type="Multiply" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>128</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="1">
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>128</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="975" name="bottleneck3_1/dim_inc/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1,1" pads_begin="0,0" pads_end="0,0" strides="1,1"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1">
+ <dim>128</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="976" name="data_add_2402524030104622374" type="Const" version="opset1">
+ <data element_type="f16" offset="58674" shape="1,128,1,1" size="256"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="977" name="bottleneck3_1/dim_inc/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_1/dim_inc/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="978" name="bottleneck3_1/add/fq_input_1" type="FakeQuantize" version="opset1">
+ <data auto_broadcast="numpy" levels="256"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1"/>
+ <port id="2"/>
+ <port id="3"/>
+ <port id="4"/>
+ </input>
+ <output>
+ <port id="5" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="979" name="bottleneck3_1/add" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_1/add" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="980" name="bottleneck3_1/fn/weights31104406281049" type="Const" version="opset1">
+ <data element_type="f32" offset="1576" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="981" name="bottleneck3_1/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_1/add" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="982" name="bottleneck3_2/add/fq_input_0" type="FakeQuantize" version="opset1">
+ <data auto_broadcast="numpy" levels="256"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1"/>
+ <port id="2"/>
+ <port id="3"/>
+ <port id="4"/>
+ </input>
+ <output>
+ <port id="5" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="983" name="4414441821747" type="Const" version="opset1">
+ <data element_type="f16" offset="58930" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="984" name="4415441920628" type="Const" version="opset1">
+ <data element_type="f16" offset="58932" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="985" name="4416442022383" type="Const" version="opset1">
+ <data element_type="f16" offset="58930" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="986" name="4417442122740" type="Const" version="opset1">
+ <data element_type="f16" offset="58932" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="987" name="4584458822818" type="Const" version="opset1">
+ <data element_type="f16" offset="58934" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="988" name="4585458922263" type="Const" version="opset1">
+ <data element_type="f16" offset="58936" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="989" name="4586459022260" type="Const" version="opset1">
+ <data element_type="f16" offset="58934" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="990" name="4587459120019" type="Const" version="opset1">
+ <data element_type="f16" offset="58936" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="991" name="5524552819980" type="Const" version="opset1">
+ <data element_type="f16" offset="58938" shape="1,32,1,1" size="64"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="992" name="5525552922809" type="Const" version="opset1">
+ <data element_type="f16" offset="59002" shape="1,32,1,1" size="64"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="993" name="5526553021516" type="Const" version="opset1">
+ <data element_type="f16" offset="58938" shape="1,32,1,1" size="64"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="994" name="5527553122677" type="Const" version="opset1">
+ <data element_type="f16" offset="59002" shape="1,32,1,1" size="64"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="995" name="bottleneck3_2/dim_red/bn/mean/Fused_Mul__copy105110221/quantized1170422995" type="Const" version="opset1">
+ <data element_type="i8" offset="59066" shape="32,128,1,1" size="4096"/>
+ <output>
+ <port id="0" precision="I8">
+ <dim>32</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="996" name="bottleneck3_2/dim_red/bn/mean/Fused_Mul__copy105110221/quantized/to_f16" type="Convert" version="opset1">
+ <data destination_type="f16"/>
+ <input>
+ <port id="0">
+ <dim>32</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="1" precision="FP16">
+ <dim>32</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="997" name="bottleneck3_2/dim_red/conv/fq_weights_1/zero_point1171720169" type="Const" version="opset1">
+ <data element_type="f16" offset="63162" shape="32,1,1,1" size="64"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="998" name="bottleneck3_2/dim_red/conv/fq_weights_1/minus_zp" type="Subtract" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>32</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="1">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>32</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="999" name="bottleneck3_2/dim_red/conv/fq_weights_1/scale1171219386" type="Const" version="opset1">
+ <data element_type="f16" offset="63226" shape="32,1,1,1" size="64"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1000" name="bottleneck3_2/dim_red/conv/fq_weights_1/mulpiply_by_scale" type="Multiply" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>32</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="1">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>32</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1001" name="bottleneck3_2/dim_red/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1,1" pads_begin="0,0" pads_end="0,0" strides="1,1"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1">
+ <dim>32</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1002" name="data_add_2403324038105321561" type="Const" version="opset1">
+ <data element_type="f16" offset="63290" shape="1,32,1,1" size="64"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1003" name="bottleneck3_2/dim_red/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_2/dim_red/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1004" name="bottleneck3_2/dim_red/fn/weights31028397161055" type="Const" version="opset1">
+ <data element_type="f32" offset="1576" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1005" name="bottleneck3_2/dim_red/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_2/dim_red/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1006" name="bottleneck3_2/inner/dw1/conv/fq_input_0" type="FakeQuantize" version="opset1">
+ <data auto_broadcast="numpy" levels="256"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="2">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="3">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="4">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="5" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1007" name="16939/value1694121888" type="Const" version="opset1">
+ <data element_type="i64" offset="43778" shape="5" size="40"/>
+ <output>
+ <port id="0" precision="I64">
+ <dim>5</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1008" name="bottleneck3_2/inner/dw1/bn/mean/Fused_Mul__copy105710224/quantized1259221393" type="Const" version="opset1">
+ <data element_type="i8" offset="63354" shape="32,1,3,3" size="288"/>
+ <output>
+ <port id="0" precision="I8">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1009" name="bottleneck3_2/inner/dw1/bn/mean/Fused_Mul__copy105710224/quantized/to_f16" type="Convert" version="opset1">
+ <data destination_type="f16"/>
+ <input>
+ <port id="0">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </input>
+ <output>
+ <port id="1" precision="FP16">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1010" name="bottleneck3_2/inner/dw1/conv/fq_weights_1/zero_point1260522299" type="Const" version="opset1">
+ <data element_type="f16" offset="63642" shape="32,1,1,1" size="64"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1011" name="bottleneck3_2/inner/dw1/conv/fq_weights_1/minus_zp" type="Subtract" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ <port id="1">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1012" name="bottleneck3_2/inner/dw1/conv/fq_weights_1/scale1260021615" type="Const" version="opset1">
+ <data element_type="f16" offset="63706" shape="32,1,1,1" size="64"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1013" name="bottleneck3_2/inner/dw1/conv/fq_weights_1/mulpiply_by_scale" type="Multiply" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ <port id="1">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1014" name="16939" type="Reshape" version="opset1">
+ <data special_zero="true"/>
+ <input>
+ <port id="0">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ <port id="1">
+ <dim>5</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1015" name="bottleneck3_2/inner/dw1/conv" type="GroupConvolution" version="opset1">
+ <data auto_pad="explicit" dilations="1,1" pads_begin="1,1" pads_end="1,1" strides="1,1"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1016" name="data_add_2404124046105922929" type="Const" version="opset1">
+ <data element_type="f16" offset="63770" shape="1,32,1,1" size="64"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1017" name="bottleneck3_2/inner/dw1/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_2/inner/dw1/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1018" name="bottleneck3_2/inner/dw1/fn/weights31188400851061" type="Const" version="opset1">
+ <data element_type="f32" offset="1576" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1019" name="bottleneck3_2/inner/dw1/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_2/inner/dw1/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1020" name="bottleneck3_2/dim_inc/conv/fq_input_0" type="FakeQuantize" version="opset1">
+ <data auto_broadcast="numpy" levels="256"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1"/>
+ <port id="2"/>
+ <port id="3"/>
+ <port id="4"/>
+ </input>
+ <output>
+ <port id="5" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1021" name="bottleneck3_2/dim_inc/bn/mean/Fused_Mul__copy106310227/quantized1352820487" type="Const" version="opset1">
+ <data element_type="i8" offset="63834" shape="128,32,1,1" size="4096"/>
+ <output>
+ <port id="0" precision="I8">
+ <dim>128</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1022" name="bottleneck3_2/dim_inc/bn/mean/Fused_Mul__copy106310227/quantized/to_f16" type="Convert" version="opset1">
+ <data destination_type="f16"/>
+ <input>
+ <port id="0">
+ <dim>128</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="1" precision="FP16">
+ <dim>128</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1023" name="bottleneck3_2/dim_inc/conv/fq_weights_1/zero_point1354119497" type="Const" version="opset1">
+ <data element_type="f16" offset="67930" shape="128,1,1,1" size="256"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1024" name="bottleneck3_2/dim_inc/conv/fq_weights_1/minus_zp" type="Subtract" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>128</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="1">
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>128</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1025" name="bottleneck3_2/dim_inc/conv/fq_weights_1/scale1353619404" type="Const" version="opset1">
+ <data element_type="f16" offset="68186" shape="128,1,1,1" size="256"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1026" name="bottleneck3_2/dim_inc/conv/fq_weights_1/mulpiply_by_scale" type="Multiply" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>128</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="1">
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>128</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1027" name="bottleneck3_2/dim_inc/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1,1" pads_begin="0,0" pads_end="0,0" strides="1,1"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1">
+ <dim>128</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1028" name="data_add_2404924054106522890" type="Const" version="opset1">
+ <data element_type="f16" offset="68442" shape="1,128,1,1" size="256"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1029" name="bottleneck3_2/dim_inc/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_2/dim_inc/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1030" name="bottleneck3_2/add/fq_input_1" type="FakeQuantize" version="opset1">
+ <data auto_broadcast="numpy" levels="256"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1"/>
+ <port id="2"/>
+ <port id="3"/>
+ <port id="4"/>
+ </input>
+ <output>
+ <port id="5" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1031" name="bottleneck3_2/add" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_2/add" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1032" name="bottleneck3_2/fn/weights31052404151068" type="Const" version="opset1">
+ <data element_type="f32" offset="1576" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1033" name="bottleneck3_2/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_2/add" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1034" name="bottleneck3_3/add/fq_input_0" type="FakeQuantize" version="opset1">
+ <data auto_broadcast="numpy" levels="256"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1"/>
+ <port id="2"/>
+ <port id="3"/>
+ <port id="4"/>
+ </input>
+ <output>
+ <port id="5" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1035" name="3414341822776" type="Const" version="opset1">
+ <data element_type="f16" offset="68698" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="1036" name="3415341921687" type="Const" version="opset1">
+ <data element_type="f16" offset="68700" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="1037" name="3416342020697" type="Const" version="opset1">
+ <data element_type="f16" offset="68698" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="1038" name="3417342121660" type="Const" version="opset1">
+ <data element_type="f16" offset="68700" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="1039" name="3044304821864" type="Const" version="opset1">
+ <data element_type="f16" offset="68702" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="1040" name="3045304922281" type="Const" version="opset1">
+ <data element_type="f16" offset="68704" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="1041" name="3046305020424" type="Const" version="opset1">
+ <data element_type="f16" offset="68702" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="1042" name="3047305119983" type="Const" version="opset1">
+ <data element_type="f16" offset="68704" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="1043" name="3764376822410" type="Const" version="opset1">
+ <data element_type="f16" offset="68706" shape="1,32,1,1" size="64"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1044" name="3765376919410" type="Const" version="opset1">
+ <data element_type="f16" offset="68770" shape="1,32,1,1" size="64"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1045" name="3766377019854" type="Const" version="opset1">
+ <data element_type="f16" offset="68706" shape="1,32,1,1" size="64"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1046" name="3767377120097" type="Const" version="opset1">
+ <data element_type="f16" offset="68770" shape="1,32,1,1" size="64"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1047" name="bottleneck3_3/dim_red/bn/mean/Fused_Mul__copy107010230/quantized1256821192" type="Const" version="opset1">
+ <data element_type="i8" offset="68834" shape="32,128,1,1" size="4096"/>
+ <output>
+ <port id="0" precision="I8">
+ <dim>32</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1048" name="bottleneck3_3/dim_red/bn/mean/Fused_Mul__copy107010230/quantized/to_f16" type="Convert" version="opset1">
+ <data destination_type="f16"/>
+ <input>
+ <port id="0">
+ <dim>32</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="1" precision="FP16">
+ <dim>32</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1049" name="bottleneck3_3/dim_red/conv/fq_weights_1/zero_point1258122926" type="Const" version="opset1">
+ <data element_type="f16" offset="72930" shape="32,1,1,1" size="64"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1050" name="bottleneck3_3/dim_red/conv/fq_weights_1/minus_zp" type="Subtract" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>32</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="1">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>32</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1051" name="bottleneck3_3/dim_red/conv/fq_weights_1/scale1257622518" type="Const" version="opset1">
+ <data element_type="f16" offset="72994" shape="32,1,1,1" size="64"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1052" name="bottleneck3_3/dim_red/conv/fq_weights_1/mulpiply_by_scale" type="Multiply" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>32</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="1">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>32</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1053" name="bottleneck3_3/dim_red/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1,1" pads_begin="0,0" pads_end="0,0" strides="1,1"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1">
+ <dim>32</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1054" name="data_add_2405724062107219899" type="Const" version="opset1">
+ <data element_type="f16" offset="73058" shape="1,32,1,1" size="64"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1055" name="bottleneck3_3/dim_red/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_3/dim_red/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1056" name="bottleneck3_3/dim_red/fn/weights30868401061074" type="Const" version="opset1">
+ <data element_type="f32" offset="1576" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1057" name="bottleneck3_3/dim_red/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_3/dim_red/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1058" name="bottleneck3_3/inner/dw1/conv/fq_input_0" type="FakeQuantize" version="opset1">
+ <data auto_broadcast="numpy" levels="256"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="2">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="3">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="4">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="5" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1059" name="16851/value1685320610" type="Const" version="opset1">
+ <data element_type="i64" offset="43778" shape="5" size="40"/>
+ <output>
+ <port id="0" precision="I64">
+ <dim>5</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1060" name="bottleneck3_3/inner/dw1/bn/mean/Fused_Mul__copy107610233/quantized1276022884" type="Const" version="opset1">
+ <data element_type="i8" offset="73122" shape="32,1,3,3" size="288"/>
+ <output>
+ <port id="0" precision="I8">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1061" name="bottleneck3_3/inner/dw1/bn/mean/Fused_Mul__copy107610233/quantized/to_f16" type="Convert" version="opset1">
+ <data destination_type="f16"/>
+ <input>
+ <port id="0">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </input>
+ <output>
+ <port id="1" precision="FP16">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1062" name="bottleneck3_3/inner/dw1/conv/fq_weights_1/zero_point1277321207" type="Const" version="opset1">
+ <data element_type="f16" offset="73410" shape="32,1,1,1" size="64"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1063" name="bottleneck3_3/inner/dw1/conv/fq_weights_1/minus_zp" type="Subtract" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ <port id="1">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1064" name="bottleneck3_3/inner/dw1/conv/fq_weights_1/scale1276820181" type="Const" version="opset1">
+ <data element_type="f16" offset="73474" shape="32,1,1,1" size="64"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1065" name="bottleneck3_3/inner/dw1/conv/fq_weights_1/mulpiply_by_scale" type="Multiply" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ <port id="1">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1066" name="16851" type="Reshape" version="opset1">
+ <data special_zero="true"/>
+ <input>
+ <port id="0">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ <port id="1">
+ <dim>5</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1067" name="bottleneck3_3/inner/dw1/conv" type="GroupConvolution" version="opset1">
+ <data auto_pad="explicit" dilations="1,1" pads_begin="1,1" pads_end="1,1" strides="1,1"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1068" name="data_add_2406524070107822515" type="Const" version="opset1">
+ <data element_type="f16" offset="73538" shape="1,32,1,1" size="64"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1069" name="bottleneck3_3/inner/dw1/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_3/inner/dw1/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1070" name="bottleneck3_3/inner/dw1/fn/weights31136405621080" type="Const" version="opset1">
+ <data element_type="f32" offset="1576" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1071" name="bottleneck3_3/inner/dw1/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_3/inner/dw1/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1072" name="bottleneck3_3/dim_inc/conv/fq_input_0" type="FakeQuantize" version="opset1">
+ <data auto_broadcast="numpy" levels="256"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1"/>
+ <port id="2"/>
+ <port id="3"/>
+ <port id="4"/>
+ </input>
+ <output>
+ <port id="5" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1073" name="bottleneck3_3/dim_inc/bn/mean/Fused_Mul__copy108210236/quantized1220820997" type="Const" version="opset1">
+ <data element_type="i8" offset="73602" shape="128,32,1,1" size="4096"/>
+ <output>
+ <port id="0" precision="I8">
+ <dim>128</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1074" name="bottleneck3_3/dim_inc/bn/mean/Fused_Mul__copy108210236/quantized/to_f16" type="Convert" version="opset1">
+ <data destination_type="f16"/>
+ <input>
+ <port id="0">
+ <dim>128</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="1" precision="FP16">
+ <dim>128</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1075" name="bottleneck3_3/dim_inc/conv/fq_weights_1/zero_point1222120646" type="Const" version="opset1">
+ <data element_type="f16" offset="77698" shape="128,1,1,1" size="256"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1076" name="bottleneck3_3/dim_inc/conv/fq_weights_1/minus_zp" type="Subtract" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>128</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="1">
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>128</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1077" name="bottleneck3_3/dim_inc/conv/fq_weights_1/scale1221622086" type="Const" version="opset1">
+ <data element_type="f16" offset="77954" shape="128,1,1,1" size="256"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1078" name="bottleneck3_3/dim_inc/conv/fq_weights_1/mulpiply_by_scale" type="Multiply" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>128</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="1">
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>128</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1079" name="bottleneck3_3/dim_inc/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1,1" pads_begin="0,0" pads_end="0,0" strides="1,1"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1">
+ <dim>128</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1080" name="data_add_2407324078108419977" type="Const" version="opset1">
+ <data element_type="f16" offset="78210" shape="1,128,1,1" size="256"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1081" name="bottleneck3_3/dim_inc/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_3/dim_inc/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1082" name="bottleneck3_3/add/fq_input_1" type="FakeQuantize" version="opset1">
+ <data auto_broadcast="numpy" levels="256"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1"/>
+ <port id="2"/>
+ <port id="3"/>
+ <port id="4"/>
+ </input>
+ <output>
+ <port id="5" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1083" name="bottleneck3_3/add" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_3/add" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1084" name="bottleneck3_3/fn/weights31140400911087" type="Const" version="opset1">
+ <data element_type="f32" offset="1576" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1085" name="bottleneck3_3/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_3/add" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1086" name="bottleneck3_4/add/fq_input_0" type="FakeQuantize" version="opset1">
+ <data auto_broadcast="numpy" levels="256"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1"/>
+ <port id="2"/>
+ <port id="3"/>
+ <port id="4"/>
+ </input>
+ <output>
+ <port id="5" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1087" name="2634263820241" type="Const" version="opset1">
+ <data element_type="f16" offset="78466" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="1088" name="2635263920337" type="Const" version="opset1">
+ <data element_type="f16" offset="78468" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="1089" name="2636264021465" type="Const" version="opset1">
+ <data element_type="f16" offset="78466" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="1090" name="2637264121453" type="Const" version="opset1">
+ <data element_type="f16" offset="78468" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="1091" name="4824482820742" type="Const" version="opset1">
+ <data element_type="f16" offset="78470" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="1092" name="4825482922785" type="Const" version="opset1">
+ <data element_type="f16" offset="78472" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="1093" name="4826483020187" type="Const" version="opset1">
+ <data element_type="f16" offset="78470" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="1094" name="4827483120166" type="Const" version="opset1">
+ <data element_type="f16" offset="78472" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="1095" name="5284528822872" type="Const" version="opset1">
+ <data element_type="f16" offset="78474" shape="1,32,1,1" size="64"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1096" name="5285528919452" type="Const" version="opset1">
+ <data element_type="f16" offset="78538" shape="1,32,1,1" size="64"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1097" name="5286529022962" type="Const" version="opset1">
+ <data element_type="f16" offset="78474" shape="1,32,1,1" size="64"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1098" name="5287529121930" type="Const" version="opset1">
+ <data element_type="f16" offset="78538" shape="1,32,1,1" size="64"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1099" name="bottleneck3_4/dim_red/bn/mean/Fused_Mul__copy108910239/quantized1319221501" type="Const" version="opset1">
+ <data element_type="i8" offset="78602" shape="32,128,1,1" size="4096"/>
+ <output>
+ <port id="0" precision="I8">
+ <dim>32</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1100" name="bottleneck3_4/dim_red/bn/mean/Fused_Mul__copy108910239/quantized/to_f16" type="Convert" version="opset1">
+ <data destination_type="f16"/>
+ <input>
+ <port id="0">
+ <dim>32</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="1" precision="FP16">
+ <dim>32</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1101" name="bottleneck3_4/dim_red/conv/fq_weights_1/zero_point1320522071" type="Const" version="opset1">
+ <data element_type="f16" offset="82698" shape="32,1,1,1" size="64"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1102" name="bottleneck3_4/dim_red/conv/fq_weights_1/minus_zp" type="Subtract" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>32</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="1">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>32</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1103" name="bottleneck3_4/dim_red/conv/fq_weights_1/scale1320020601" type="Const" version="opset1">
+ <data element_type="f16" offset="82762" shape="32,1,1,1" size="64"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1104" name="bottleneck3_4/dim_red/conv/fq_weights_1/mulpiply_by_scale" type="Multiply" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>32</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="1">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>32</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1105" name="bottleneck3_4/dim_red/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1,1" pads_begin="0,0" pads_end="0,0" strides="1,1"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1">
+ <dim>32</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1106" name="data_add_2408124086109121555" type="Const" version="opset1">
+ <data element_type="f16" offset="82826" shape="1,32,1,1" size="64"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1107" name="bottleneck3_4/dim_red/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_4/dim_red/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1108" name="bottleneck3_4/dim_red/fn/weights31152397191093" type="Const" version="opset1">
+ <data element_type="f32" offset="1576" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1109" name="bottleneck3_4/dim_red/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_4/dim_red/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1110" name="bottleneck3_4/inner/dw1/conv/fq_input_0" type="FakeQuantize" version="opset1">
+ <data auto_broadcast="numpy" levels="256"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="2">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="3">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="4">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="5" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1111" name="16927/value1692920505" type="Const" version="opset1">
+ <data element_type="i64" offset="43778" shape="5" size="40"/>
+ <output>
+ <port id="0" precision="I64">
+ <dim>5</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1112" name="bottleneck3_4/inner/dw1/bn/mean/Fused_Mul__copy109510242/quantized1165622575" type="Const" version="opset1">
+ <data element_type="i8" offset="82890" shape="32,1,3,3" size="288"/>
+ <output>
+ <port id="0" precision="I8">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1113" name="bottleneck3_4/inner/dw1/bn/mean/Fused_Mul__copy109510242/quantized/to_f16" type="Convert" version="opset1">
+ <data destination_type="f16"/>
+ <input>
+ <port id="0">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </input>
+ <output>
+ <port id="1" precision="FP16">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1114" name="bottleneck3_4/inner/dw1/conv/fq_weights_1/zero_point1166921306" type="Const" version="opset1">
+ <data element_type="f16" offset="83178" shape="32,1,1,1" size="64"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1115" name="bottleneck3_4/inner/dw1/conv/fq_weights_1/minus_zp" type="Subtract" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ <port id="1">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1116" name="bottleneck3_4/inner/dw1/conv/fq_weights_1/scale1166421477" type="Const" version="opset1">
+ <data element_type="f16" offset="83242" shape="32,1,1,1" size="64"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1117" name="bottleneck3_4/inner/dw1/conv/fq_weights_1/mulpiply_by_scale" type="Multiply" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ <port id="1">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1118" name="16927" type="Reshape" version="opset1">
+ <data special_zero="true"/>
+ <input>
+ <port id="0">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ <port id="1">
+ <dim>5</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1119" name="bottleneck3_4/inner/dw1/conv" type="GroupConvolution" version="opset1">
+ <data auto_pad="explicit" dilations="1,1" pads_begin="1,1" pads_end="1,1" strides="1,1"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1120" name="data_add_2408924094109719548" type="Const" version="opset1">
+ <data element_type="f16" offset="83306" shape="1,32,1,1" size="64"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1121" name="bottleneck3_4/inner/dw1/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_4/inner/dw1/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1122" name="bottleneck3_4/inner/dw1/fn/weights30980405111099" type="Const" version="opset1">
+ <data element_type="f32" offset="1576" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1123" name="bottleneck3_4/inner/dw1/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_4/inner/dw1/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1124" name="bottleneck3_4/dim_inc/conv/fq_input_0" type="FakeQuantize" version="opset1">
+ <data auto_broadcast="numpy" levels="256"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1"/>
+ <port id="2"/>
+ <port id="3"/>
+ <port id="4"/>
+ </input>
+ <output>
+ <port id="5" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1125" name="bottleneck3_4/dim_inc/bn/mean/Fused_Mul__copy110110245/quantized1204021180" type="Const" version="opset1">
+ <data element_type="i8" offset="83370" shape="128,32,1,1" size="4096"/>
+ <output>
+ <port id="0" precision="I8">
+ <dim>128</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1126" name="bottleneck3_4/dim_inc/bn/mean/Fused_Mul__copy110110245/quantized/to_f16" type="Convert" version="opset1">
+ <data destination_type="f16"/>
+ <input>
+ <port id="0">
+ <dim>128</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="1" precision="FP16">
+ <dim>128</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1127" name="bottleneck3_4/dim_inc/conv/fq_weights_1/zero_point1205320961" type="Const" version="opset1">
+ <data element_type="f16" offset="87466" shape="128,1,1,1" size="256"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1128" name="bottleneck3_4/dim_inc/conv/fq_weights_1/minus_zp" type="Subtract" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>128</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="1">
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>128</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1129" name="bottleneck3_4/dim_inc/conv/fq_weights_1/scale1204821165" type="Const" version="opset1">
+ <data element_type="f16" offset="87722" shape="128,1,1,1" size="256"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1130" name="bottleneck3_4/dim_inc/conv/fq_weights_1/mulpiply_by_scale" type="Multiply" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>128</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="1">
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>128</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1131" name="bottleneck3_4/dim_inc/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1,1" pads_begin="0,0" pads_end="0,0" strides="1,1"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1">
+ <dim>128</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1132" name="data_add_2409724102110321096" type="Const" version="opset1">
+ <data element_type="f16" offset="87978" shape="1,128,1,1" size="256"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1133" name="bottleneck3_4/dim_inc/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_4/dim_inc/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1134" name="bottleneck3_4/add/fq_input_1" type="FakeQuantize" version="opset1">
+ <data auto_broadcast="numpy" levels="256"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1"/>
+ <port id="2"/>
+ <port id="3"/>
+ <port id="4"/>
+ </input>
+ <output>
+ <port id="5" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1135" name="bottleneck3_4/add" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_4/add" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1136" name="bottleneck3_4/fn/weights30792399381106" type="Const" version="opset1">
+ <data element_type="f32" offset="1576" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1137" name="bottleneck3_4/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_4/add" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1138" name="bottleneck3_5/add/fq_input_0" type="FakeQuantize" version="opset1">
+ <data auto_broadcast="numpy" levels="256"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1"/>
+ <port id="2"/>
+ <port id="3"/>
+ <port id="4"/>
+ </input>
+ <output>
+ <port id="5" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1139" name="4034403822017" type="Const" version="opset1">
+ <data element_type="f16" offset="88234" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="1140" name="4035403922605" type="Const" version="opset1">
+ <data element_type="f16" offset="88236" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="1141" name="4036404022398" type="Const" version="opset1">
+ <data element_type="f16" offset="88234" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="1142" name="4037404119542" type="Const" version="opset1">
+ <data element_type="f16" offset="88236" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="1143" name="2924292820574" type="Const" version="opset1">
+ <data element_type="f16" offset="88238" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="1144" name="2925292921171" type="Const" version="opset1">
+ <data element_type="f16" offset="88240" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="1145" name="2926293020640" type="Const" version="opset1">
+ <data element_type="f16" offset="88238" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="1146" name="2927293122992" type="Const" version="opset1">
+ <data element_type="f16" offset="88240" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="1147" name="5064506820307" type="Const" version="opset1">
+ <data element_type="f16" offset="88242" shape="1,32,1,1" size="64"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1148" name="5065506919701" type="Const" version="opset1">
+ <data element_type="f16" offset="88306" shape="1,32,1,1" size="64"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1149" name="5066507019914" type="Const" version="opset1">
+ <data element_type="f16" offset="88242" shape="1,32,1,1" size="64"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1150" name="5067507122128" type="Const" version="opset1">
+ <data element_type="f16" offset="88306" shape="1,32,1,1" size="64"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1151" name="bottleneck3_5/dim_red/bn/mean/Fused_Mul__copy110810248/quantized1405620034" type="Const" version="opset1">
+ <data element_type="i8" offset="88370" shape="32,128,1,1" size="4096"/>
+ <output>
+ <port id="0" precision="I8">
+ <dim>32</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1152" name="bottleneck3_5/dim_red/bn/mean/Fused_Mul__copy110810248/quantized/to_f16" type="Convert" version="opset1">
+ <data destination_type="f16"/>
+ <input>
+ <port id="0">
+ <dim>32</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="1" precision="FP16">
+ <dim>32</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1153" name="bottleneck3_5/dim_red/conv/fq_weights_1/zero_point1406921690" type="Const" version="opset1">
+ <data element_type="f16" offset="92466" shape="32,1,1,1" size="64"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1154" name="bottleneck3_5/dim_red/conv/fq_weights_1/minus_zp" type="Subtract" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>32</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="1">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>32</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1155" name="bottleneck3_5/dim_red/conv/fq_weights_1/scale1406419581" type="Const" version="opset1">
+ <data element_type="f16" offset="92530" shape="32,1,1,1" size="64"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1156" name="bottleneck3_5/dim_red/conv/fq_weights_1/mulpiply_by_scale" type="Multiply" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>32</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="1">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>32</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1157" name="bottleneck3_5/dim_red/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1,1" pads_begin="0,0" pads_end="0,0" strides="1,1"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1">
+ <dim>32</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1158" name="data_add_2410524110111019524" type="Const" version="opset1">
+ <data element_type="f16" offset="92594" shape="1,32,1,1" size="64"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1159" name="bottleneck3_5/dim_red/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_5/dim_red/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1160" name="bottleneck3_5/dim_red/fn/weights30788405831112" type="Const" version="opset1">
+ <data element_type="f32" offset="1576" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1161" name="bottleneck3_5/dim_red/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_5/dim_red/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1162" name="bottleneck3_5/inner/dw1/conv/fq_input_0" type="FakeQuantize" version="opset1">
+ <data auto_broadcast="numpy" levels="256"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="2">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="3">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="4">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="5" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1163" name="16919/value1692120664" type="Const" version="opset1">
+ <data element_type="i64" offset="43778" shape="5" size="40"/>
+ <output>
+ <port id="0" precision="I64">
+ <dim>5</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1164" name="bottleneck3_5/inner/dw1/bn/mean/Fused_Mul__copy111410251/quantized1218421648" type="Const" version="opset1">
+ <data element_type="i8" offset="92658" shape="32,1,3,3" size="288"/>
+ <output>
+ <port id="0" precision="I8">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1165" name="bottleneck3_5/inner/dw1/bn/mean/Fused_Mul__copy111410251/quantized/to_f16" type="Convert" version="opset1">
+ <data destination_type="f16"/>
+ <input>
+ <port id="0">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </input>
+ <output>
+ <port id="1" precision="FP16">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1166" name="bottleneck3_5/inner/dw1/conv/fq_weights_1/zero_point1219720637" type="Const" version="opset1">
+ <data element_type="f16" offset="92946" shape="32,1,1,1" size="64"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1167" name="bottleneck3_5/inner/dw1/conv/fq_weights_1/minus_zp" type="Subtract" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ <port id="1">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1168" name="bottleneck3_5/inner/dw1/conv/fq_weights_1/scale1219221819" type="Const" version="opset1">
+ <data element_type="f16" offset="93010" shape="32,1,1,1" size="64"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1169" name="bottleneck3_5/inner/dw1/conv/fq_weights_1/mulpiply_by_scale" type="Multiply" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ <port id="1">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1170" name="16919" type="Reshape" version="opset1">
+ <data special_zero="true"/>
+ <input>
+ <port id="0">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ <port id="1">
+ <dim>5</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1171" name="bottleneck3_5/inner/dw1/conv" type="GroupConvolution" version="opset1">
+ <data auto_pad="explicit" dilations="1,1" pads_begin="1,1" pads_end="1,1" strides="1,1"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1172" name="data_add_2411324118111621126" type="Const" version="opset1">
+ <data element_type="f16" offset="93074" shape="1,32,1,1" size="64"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1173" name="bottleneck3_5/inner/dw1/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_5/inner/dw1/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1174" name="bottleneck3_5/inner/dw1/fn/weights30796403941118" type="Const" version="opset1">
+ <data element_type="f32" offset="1576" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1175" name="bottleneck3_5/inner/dw1/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_5/inner/dw1/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1176" name="bottleneck3_5/dim_inc/conv/fq_input_0" type="FakeQuantize" version="opset1">
+ <data auto_broadcast="numpy" levels="256"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1"/>
+ <port id="2"/>
+ <port id="3"/>
+ <port id="4"/>
+ </input>
+ <output>
+ <port id="5" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1177" name="bottleneck3_5/dim_inc/bn/mean/Fused_Mul__copy112010254/quantized1338420571" type="Const" version="opset1">
+ <data element_type="i8" offset="93138" shape="128,32,1,1" size="4096"/>
+ <output>
+ <port id="0" precision="I8">
+ <dim>128</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1178" name="bottleneck3_5/dim_inc/bn/mean/Fused_Mul__copy112010254/quantized/to_f16" type="Convert" version="opset1">
+ <data destination_type="f16"/>
+ <input>
+ <port id="0">
+ <dim>128</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="1" precision="FP16">
+ <dim>128</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1179" name="bottleneck3_5/dim_inc/conv/fq_weights_1/zero_point1339721069" type="Const" version="opset1">
+ <data element_type="f16" offset="97234" shape="128,1,1,1" size="256"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1180" name="bottleneck3_5/dim_inc/conv/fq_weights_1/minus_zp" type="Subtract" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>128</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="1">
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>128</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1181" name="bottleneck3_5/dim_inc/conv/fq_weights_1/scale1339219731" type="Const" version="opset1">
+ <data element_type="f16" offset="97490" shape="128,1,1,1" size="256"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1182" name="bottleneck3_5/dim_inc/conv/fq_weights_1/mulpiply_by_scale" type="Multiply" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>128</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="1">
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>128</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1183" name="bottleneck3_5/dim_inc/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1,1" pads_begin="0,0" pads_end="0,0" strides="1,1"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1">
+ <dim>128</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1184" name="data_add_2412124126112222830" type="Const" version="opset1">
+ <data element_type="f16" offset="97746" shape="1,128,1,1" size="256"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1185" name="bottleneck3_5/dim_inc/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_5/dim_inc/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1186" name="bottleneck3_5/add/fq_input_1" type="FakeQuantize" version="opset1">
+ <data auto_broadcast="numpy" levels="256"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1"/>
+ <port id="2"/>
+ <port id="3"/>
+ <port id="4"/>
+ </input>
+ <output>
+ <port id="5" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1187" name="bottleneck3_5/add" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_5/add" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1188" name="bottleneck3_5/fn/weights30828406371125" type="Const" version="opset1">
+ <data element_type="f32" offset="1576" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1189" name="bottleneck3_5/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_5/add" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1190" name="bottleneck3_6/add/fq_input_0" type="FakeQuantize" version="opset1">
+ <data auto_broadcast="numpy" levels="256"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1"/>
+ <port id="2"/>
+ <port id="3"/>
+ <port id="4"/>
+ </input>
+ <output>
+ <port id="5" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1191" name="3834383822431" type="Const" version="opset1">
+ <data element_type="f16" offset="98002" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="1192" name="3835383921294" type="Const" version="opset1">
+ <data element_type="f16" offset="98004" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="1193" name="3836384019779" type="Const" version="opset1">
+ <data element_type="f16" offset="98002" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="1194" name="3837384121441" type="Const" version="opset1">
+ <data element_type="f16" offset="98004" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="1195" name="3464346820745" type="Const" version="opset1">
+ <data element_type="f16" offset="98006" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="1196" name="3465346922710" type="Const" version="opset1">
+ <data element_type="f16" offset="98008" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="1197" name="3466347022125" type="Const" version="opset1">
+ <data element_type="f16" offset="98006" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="1198" name="3467347119662" type="Const" version="opset1">
+ <data element_type="f16" offset="98008" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="1199" name="3344334819866" type="Const" version="opset1">
+ <data element_type="f16" offset="98010" shape="1,32,1,1" size="64"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1200" name="3345334922596" type="Const" version="opset1">
+ <data element_type="f16" offset="98074" shape="1,32,1,1" size="64"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1201" name="3346335022317" type="Const" version="opset1">
+ <data element_type="f16" offset="98010" shape="1,32,1,1" size="64"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1202" name="3347335119845" type="Const" version="opset1">
+ <data element_type="f16" offset="98074" shape="1,32,1,1" size="64"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1203" name="bottleneck3_6/dim_red/bn/mean/Fused_Mul__copy112710257/quantized1295222950" type="Const" version="opset1">
+ <data element_type="i8" offset="98138" shape="32,128,1,1" size="4096"/>
+ <output>
+ <port id="0" precision="I8">
+ <dim>32</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1204" name="bottleneck3_6/dim_red/bn/mean/Fused_Mul__copy112710257/quantized/to_f16" type="Convert" version="opset1">
+ <data destination_type="f16"/>
+ <input>
+ <port id="0">
+ <dim>32</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="1" precision="FP16">
+ <dim>32</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1205" name="bottleneck3_6/dim_red/conv/fq_weights_1/zero_point1296520625" type="Const" version="opset1">
+ <data element_type="f16" offset="102234" shape="32,1,1,1" size="64"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1206" name="bottleneck3_6/dim_red/conv/fq_weights_1/minus_zp" type="Subtract" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>32</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="1">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>32</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1207" name="bottleneck3_6/dim_red/conv/fq_weights_1/scale1296020553" type="Const" version="opset1">
+ <data element_type="f16" offset="102298" shape="32,1,1,1" size="64"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1208" name="bottleneck3_6/dim_red/conv/fq_weights_1/mulpiply_by_scale" type="Multiply" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>32</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="1">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>32</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1209" name="bottleneck3_6/dim_red/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1,1" pads_begin="0,0" pads_end="0,0" strides="1,1"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1">
+ <dim>32</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1210" name="data_add_2412924134112922899" type="Const" version="opset1">
+ <data element_type="f16" offset="102362" shape="1,32,1,1" size="64"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1211" name="bottleneck3_6/dim_red/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_6/dim_red/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1212" name="bottleneck3_6/dim_red/fn/weights30772406221131" type="Const" version="opset1">
+ <data element_type="f32" offset="1576" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1213" name="bottleneck3_6/dim_red/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_6/dim_red/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1214" name="bottleneck3_6/inner/dw1/conv/fq_input_0" type="FakeQuantize" version="opset1">
+ <data auto_broadcast="numpy" levels="256"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="2">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="3">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="4">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="5" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1215" name="16831/value1683321768" type="Const" version="opset1">
+ <data element_type="i64" offset="43778" shape="5" size="40"/>
+ <output>
+ <port id="0" precision="I64">
+ <dim>5</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1216" name="bottleneck3_6/inner/dw1/bn/mean/Fused_Mul__copy113310260/quantized1252022770" type="Const" version="opset1">
+ <data element_type="i8" offset="102426" shape="32,1,3,3" size="288"/>
+ <output>
+ <port id="0" precision="I8">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1217" name="bottleneck3_6/inner/dw1/bn/mean/Fused_Mul__copy113310260/quantized/to_f16" type="Convert" version="opset1">
+ <data destination_type="f16"/>
+ <input>
+ <port id="0">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </input>
+ <output>
+ <port id="1" precision="FP16">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1218" name="bottleneck3_6/inner/dw1/conv/fq_weights_1/zero_point1253322650" type="Const" version="opset1">
+ <data element_type="f16" offset="102714" shape="32,1,1,1" size="64"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1219" name="bottleneck3_6/inner/dw1/conv/fq_weights_1/minus_zp" type="Subtract" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ <port id="1">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1220" name="bottleneck3_6/inner/dw1/conv/fq_weights_1/scale1252821573" type="Const" version="opset1">
+ <data element_type="f16" offset="102778" shape="32,1,1,1" size="64"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1221" name="bottleneck3_6/inner/dw1/conv/fq_weights_1/mulpiply_by_scale" type="Multiply" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ <port id="1">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1222" name="16831" type="Reshape" version="opset1">
+ <data special_zero="true"/>
+ <input>
+ <port id="0">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ <port id="1">
+ <dim>5</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1223" name="bottleneck3_6/inner/dw1/conv" type="GroupConvolution" version="opset1">
+ <data auto_pad="explicit" dilations="1,1" pads_begin="1,1" pads_end="1,1" strides="1,1"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1224" name="data_add_2413724142113520937" type="Const" version="opset1">
+ <data element_type="f16" offset="102842" shape="1,32,1,1" size="64"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1225" name="bottleneck3_6/inner/dw1/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_6/inner/dw1/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1226" name="bottleneck3_6/inner/dw1/fn/weights31124397131137" type="Const" version="opset1">
+ <data element_type="f32" offset="1576" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1227" name="bottleneck3_6/inner/dw1/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_6/inner/dw1/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1228" name="bottleneck3_6/dim_inc/conv/fq_input_0" type="FakeQuantize" version="opset1">
+ <data auto_broadcast="numpy" levels="256"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1"/>
+ <port id="2"/>
+ <port id="3"/>
+ <port id="4"/>
+ </input>
+ <output>
+ <port id="5" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1229" name="bottleneck3_6/dim_inc/bn/mean/Fused_Mul__copy113910263/quantized1336021981" type="Const" version="opset1">
+ <data element_type="i8" offset="102906" shape="128,32,1,1" size="4096"/>
+ <output>
+ <port id="0" precision="I8">
+ <dim>128</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1230" name="bottleneck3_6/dim_inc/bn/mean/Fused_Mul__copy113910263/quantized/to_f16" type="Convert" version="opset1">
+ <data destination_type="f16"/>
+ <input>
+ <port id="0">
+ <dim>128</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="1" precision="FP16">
+ <dim>128</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1231" name="bottleneck3_6/dim_inc/conv/fq_weights_1/zero_point1337322548" type="Const" version="opset1">
+ <data element_type="f16" offset="107002" shape="128,1,1,1" size="256"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1232" name="bottleneck3_6/dim_inc/conv/fq_weights_1/minus_zp" type="Subtract" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>128</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="1">
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>128</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1233" name="bottleneck3_6/dim_inc/conv/fq_weights_1/scale1336821696" type="Const" version="opset1">
+ <data element_type="f16" offset="107258" shape="128,1,1,1" size="256"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1234" name="bottleneck3_6/dim_inc/conv/fq_weights_1/mulpiply_by_scale" type="Multiply" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>128</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="1">
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>128</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1235" name="bottleneck3_6/dim_inc/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1,1" pads_begin="0,0" pads_end="0,0" strides="1,1"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1">
+ <dim>128</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1236" name="data_add_2414524150114120967" type="Const" version="opset1">
+ <data element_type="f16" offset="107514" shape="1,128,1,1" size="256"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1237" name="bottleneck3_6/dim_inc/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_6/dim_inc/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1238" name="bottleneck3_6/add/fq_input_1" type="FakeQuantize" version="opset1">
+ <data auto_broadcast="numpy" levels="256"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1"/>
+ <port id="2"/>
+ <port id="3"/>
+ <port id="4"/>
+ </input>
+ <output>
+ <port id="5" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1239" name="bottleneck3_6/add" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_6/add" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1240" name="bottleneck3_6/fn/weights31032396591144" type="Const" version="opset1">
+ <data element_type="f32" offset="1576" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1241" name="bottleneck3_6/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_6/add" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1242" name="bottleneck3_7/add/fq_input_0" type="FakeQuantize" version="opset1">
+ <data auto_broadcast="numpy" levels="256"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1"/>
+ <port id="2"/>
+ <port id="3"/>
+ <port id="4"/>
+ </input>
+ <output>
+ <port id="5" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1243" name="4674467820403" type="Const" version="opset1">
+ <data element_type="f16" offset="107770" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="1244" name="4675467921108" type="Const" version="opset1">
+ <data element_type="f16" offset="107772" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="1245" name="4676468022062" type="Const" version="opset1">
+ <data element_type="f16" offset="107770" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="1246" name="4677468119893" type="Const" version="opset1">
+ <data element_type="f16" offset="107772" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="1247" name="4344434821423" type="Const" version="opset1">
+ <data element_type="f16" offset="107774" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="1248" name="4345434919683" type="Const" version="opset1">
+ <data element_type="f16" offset="107776" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="1249" name="4346435022674" type="Const" version="opset1">
+ <data element_type="f16" offset="107774" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="1250" name="4347435121381" type="Const" version="opset1">
+ <data element_type="f16" offset="107776" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="1251" name="4744474821639" type="Const" version="opset1">
+ <data element_type="f16" offset="107778" shape="1,32,1,1" size="64"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1252" name="4745474921519" type="Const" version="opset1">
+ <data element_type="f16" offset="107842" shape="1,32,1,1" size="64"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1253" name="4746475020751" type="Const" version="opset1">
+ <data element_type="f16" offset="107778" shape="1,32,1,1" size="64"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1254" name="4747475120016" type="Const" version="opset1">
+ <data element_type="f16" offset="107842" shape="1,32,1,1" size="64"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1255" name="bottleneck3_7/dim_red/bn/mean/Fused_Mul__copy114610266/quantized1254419824" type="Const" version="opset1">
+ <data element_type="i8" offset="107906" shape="32,128,1,1" size="4096"/>
+ <output>
+ <port id="0" precision="I8">
+ <dim>32</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1256" name="bottleneck3_7/dim_red/bn/mean/Fused_Mul__copy114610266/quantized/to_f16" type="Convert" version="opset1">
+ <data destination_type="f16"/>
+ <input>
+ <port id="0">
+ <dim>32</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="1" precision="FP16">
+ <dim>32</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1257" name="bottleneck3_7/dim_red/conv/fq_weights_1/zero_point1255722533" type="Const" version="opset1">
+ <data element_type="f16" offset="112002" shape="32,1,1,1" size="64"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1258" name="bottleneck3_7/dim_red/conv/fq_weights_1/minus_zp" type="Subtract" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>32</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="1">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>32</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1259" name="bottleneck3_7/dim_red/conv/fq_weights_1/scale1255220076" type="Const" version="opset1">
+ <data element_type="f16" offset="112066" shape="32,1,1,1" size="64"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1260" name="bottleneck3_7/dim_red/conv/fq_weights_1/mulpiply_by_scale" type="Multiply" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>32</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="1">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>32</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1261" name="bottleneck3_7/dim_red/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1,1" pads_begin="0,0" pads_end="0,0" strides="1,1"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1">
+ <dim>32</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1262" name="data_add_2415324158114819506" type="Const" version="opset1">
+ <data element_type="f16" offset="112130" shape="1,32,1,1" size="64"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1263" name="bottleneck3_7/dim_red/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_7/dim_red/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1264" name="bottleneck3_7/dim_red/fn/weights30988406521150" type="Const" version="opset1">
+ <data element_type="f32" offset="1576" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1265" name="bottleneck3_7/dim_red/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_7/dim_red/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1266" name="bottleneck3_7/inner/dw1/conv/fq_input_0" type="FakeQuantize" version="opset1">
+ <data auto_broadcast="numpy" levels="256"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="2">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="3">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="4">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="5" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1267" name="16899/value1690122560" type="Const" version="opset1">
+ <data element_type="i64" offset="43778" shape="5" size="40"/>
+ <output>
+ <port id="0" precision="I64">
+ <dim>5</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1268" name="bottleneck3_7/inner/dw1/bn/mean/Fused_Mul__copy115210269/quantized1148821078" type="Const" version="opset1">
+ <data element_type="i8" offset="112194" shape="32,1,3,3" size="288"/>
+ <output>
+ <port id="0" precision="I8">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1269" name="bottleneck3_7/inner/dw1/bn/mean/Fused_Mul__copy115210269/quantized/to_f16" type="Convert" version="opset1">
+ <data destination_type="f16"/>
+ <input>
+ <port id="0">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </input>
+ <output>
+ <port id="1" precision="FP16">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1270" name="bottleneck3_7/inner/dw1/conv/fq_weights_1/zero_point1150120268" type="Const" version="opset1">
+ <data element_type="f16" offset="112482" shape="32,1,1,1" size="64"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1271" name="bottleneck3_7/inner/dw1/conv/fq_weights_1/minus_zp" type="Subtract" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ <port id="1">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1272" name="bottleneck3_7/inner/dw1/conv/fq_weights_1/scale1149622107" type="Const" version="opset1">
+ <data element_type="f16" offset="112546" shape="32,1,1,1" size="64"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1273" name="bottleneck3_7/inner/dw1/conv/fq_weights_1/mulpiply_by_scale" type="Multiply" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ <port id="1">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1274" name="16899" type="Reshape" version="opset1">
+ <data special_zero="true"/>
+ <input>
+ <port id="0">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ <port id="1">
+ <dim>5</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1275" name="bottleneck3_7/inner/dw1/conv" type="GroupConvolution" version="opset1">
+ <data auto_pad="explicit" dilations="1,1" pads_begin="1,1" pads_end="1,1" strides="1,1"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1276" name="data_add_2416124166115421378" type="Const" version="opset1">
+ <data element_type="f16" offset="112610" shape="1,32,1,1" size="64"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1277" name="bottleneck3_7/inner/dw1/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_7/inner/dw1/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1278" name="bottleneck3_7/inner/dw1/fn/weights31064397071156" type="Const" version="opset1">
+ <data element_type="f32" offset="1576" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1279" name="bottleneck3_7/inner/dw1/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_7/inner/dw1/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1280" name="bottleneck3_7/dim_inc/conv/fq_input_0" type="FakeQuantize" version="opset1">
+ <data auto_broadcast="numpy" levels="256"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1"/>
+ <port id="2"/>
+ <port id="3"/>
+ <port id="4"/>
+ </input>
+ <output>
+ <port id="5" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1281" name="bottleneck3_7/dim_inc/bn/mean/Fused_Mul__copy115810272/quantized1240021735" type="Const" version="opset1">
+ <data element_type="i8" offset="112674" shape="128,32,1,1" size="4096"/>
+ <output>
+ <port id="0" precision="I8">
+ <dim>128</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1282" name="bottleneck3_7/dim_inc/bn/mean/Fused_Mul__copy115810272/quantized/to_f16" type="Convert" version="opset1">
+ <data destination_type="f16"/>
+ <input>
+ <port id="0">
+ <dim>128</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="1" precision="FP16">
+ <dim>128</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1283" name="bottleneck3_7/dim_inc/conv/fq_weights_1/zero_point1241322959" type="Const" version="opset1">
+ <data element_type="f16" offset="116770" shape="128,1,1,1" size="256"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1284" name="bottleneck3_7/dim_inc/conv/fq_weights_1/minus_zp" type="Subtract" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>128</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="1">
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>128</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1285" name="bottleneck3_7/dim_inc/conv/fq_weights_1/scale1240820940" type="Const" version="opset1">
+ <data element_type="f16" offset="117026" shape="128,1,1,1" size="256"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1286" name="bottleneck3_7/dim_inc/conv/fq_weights_1/mulpiply_by_scale" type="Multiply" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>128</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="1">
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>128</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1287" name="bottleneck3_7/dim_inc/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1,1" pads_begin="0,0" pads_end="0,0" strides="1,1"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1">
+ <dim>128</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1288" name="data_add_2416924174116020970" type="Const" version="opset1">
+ <data element_type="f16" offset="117282" shape="1,128,1,1" size="256"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1289" name="bottleneck3_7/dim_inc/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_7/dim_inc/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1290" name="bottleneck3_7/add/fq_input_1" type="FakeQuantize" version="opset1">
+ <data auto_broadcast="numpy" levels="256"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1"/>
+ <port id="2"/>
+ <port id="3"/>
+ <port id="4"/>
+ </input>
+ <output>
+ <port id="5" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1291" name="bottleneck3_7/add" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_7/add" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1292" name="bottleneck3_7/fn/weights30912398481163" type="Const" version="opset1">
+ <data element_type="f32" offset="1576" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1293" name="bottleneck3_7/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_7/add" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1294" name="bottleneck3_8/add/fq_input_0" type="FakeQuantize" version="opset1">
+ <data auto_broadcast="numpy" levels="256"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1"/>
+ <port id="2"/>
+ <port id="3"/>
+ <port id="4"/>
+ </input>
+ <output>
+ <port id="5" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1295" name="4794479819713" type="Const" version="opset1">
+ <data element_type="f16" offset="117538" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="1296" name="4795479920007" type="Const" version="opset1">
+ <data element_type="f16" offset="117540" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="1297" name="4796480019725" type="Const" version="opset1">
+ <data element_type="f16" offset="117538" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="1298" name="4797480122722" type="Const" version="opset1">
+ <data element_type="f16" offset="117540" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="1299" name="3504350821351" type="Const" version="opset1">
+ <data element_type="f16" offset="117542" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="1300" name="3505350919455" type="Const" version="opset1">
+ <data element_type="f16" offset="117544" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="1301" name="3506351020973" type="Const" version="opset1">
+ <data element_type="f16" offset="117542" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="1302" name="3507351122941" type="Const" version="opset1">
+ <data element_type="f16" offset="117544" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="1303" name="4844484822203" type="Const" version="opset1">
+ <data element_type="f16" offset="117546" shape="1,32,1,1" size="64"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1304" name="4845484920283" type="Const" version="opset1">
+ <data element_type="f16" offset="117610" shape="1,32,1,1" size="64"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1305" name="4846485021891" type="Const" version="opset1">
+ <data element_type="f16" offset="117546" shape="1,32,1,1" size="64"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1306" name="4847485121924" type="Const" version="opset1">
+ <data element_type="f16" offset="117610" shape="1,32,1,1" size="64"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1307" name="bottleneck3_8/dim_red/bn/mean/Fused_Mul__copy116510275/quantized1333620232" type="Const" version="opset1">
+ <data element_type="i8" offset="117674" shape="32,128,1,1" size="4096"/>
+ <output>
+ <port id="0" precision="I8">
+ <dim>32</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1308" name="bottleneck3_8/dim_red/bn/mean/Fused_Mul__copy116510275/quantized/to_f16" type="Convert" version="opset1">
+ <data destination_type="f16"/>
+ <input>
+ <port id="0">
+ <dim>32</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="1" precision="FP16">
+ <dim>32</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1309" name="bottleneck3_8/dim_red/conv/fq_weights_1/zero_point1334919833" type="Const" version="opset1">
+ <data element_type="f16" offset="121770" shape="32,1,1,1" size="64"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1310" name="bottleneck3_8/dim_red/conv/fq_weights_1/minus_zp" type="Subtract" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>32</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="1">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>32</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1311" name="bottleneck3_8/dim_red/conv/fq_weights_1/scale1334422725" type="Const" version="opset1">
+ <data element_type="f16" offset="121834" shape="32,1,1,1" size="64"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1312" name="bottleneck3_8/dim_red/conv/fq_weights_1/mulpiply_by_scale" type="Multiply" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>32</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="1">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>32</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1313" name="bottleneck3_8/dim_red/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1,1" pads_begin="0,0" pads_end="0,0" strides="1,1"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1">
+ <dim>32</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1314" name="data_add_2417724182116722404" type="Const" version="opset1">
+ <data element_type="f16" offset="121898" shape="1,32,1,1" size="64"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1315" name="bottleneck3_8/dim_red/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_8/dim_red/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1316" name="bottleneck3_8/dim_red/fn/weights30896401691169" type="Const" version="opset1">
+ <data element_type="f32" offset="1576" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1317" name="bottleneck3_8/dim_red/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_8/dim_red/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1318" name="bottleneck3_8/inner/dw1/conv/fq_input_0" type="FakeQuantize" version="opset1">
+ <data auto_broadcast="numpy" levels="256"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="2">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="3">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="4">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="5" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1319" name="16907/value1690921534" type="Const" version="opset1">
+ <data element_type="i64" offset="43778" shape="5" size="40"/>
+ <output>
+ <port id="0" precision="I64">
+ <dim>5</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1320" name="bottleneck3_8/inner/dw1/bn/mean/Fused_Mul__copy117110278/quantized1160820304" type="Const" version="opset1">
+ <data element_type="i8" offset="121962" shape="32,1,3,3" size="288"/>
+ <output>
+ <port id="0" precision="I8">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1321" name="bottleneck3_8/inner/dw1/bn/mean/Fused_Mul__copy117110278/quantized/to_f16" type="Convert" version="opset1">
+ <data destination_type="f16"/>
+ <input>
+ <port id="0">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </input>
+ <output>
+ <port id="1" precision="FP16">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1322" name="bottleneck3_8/inner/dw1/conv/fq_weights_1/zero_point1162120820" type="Const" version="opset1">
+ <data element_type="f16" offset="122250" shape="32,1,1,1" size="64"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1323" name="bottleneck3_8/inner/dw1/conv/fq_weights_1/minus_zp" type="Subtract" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ <port id="1">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1324" name="bottleneck3_8/inner/dw1/conv/fq_weights_1/scale1161622749" type="Const" version="opset1">
+ <data element_type="f16" offset="122314" shape="32,1,1,1" size="64"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1325" name="bottleneck3_8/inner/dw1/conv/fq_weights_1/mulpiply_by_scale" type="Multiply" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ <port id="1">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1326" name="16907" type="Reshape" version="opset1">
+ <data special_zero="true"/>
+ <input>
+ <port id="0">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ <port id="1">
+ <dim>5</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1327" name="bottleneck3_8/inner/dw1/conv" type="GroupConvolution" version="opset1">
+ <data auto_pad="explicit" dilations="1,1" pads_begin="1,1" pads_end="1,1" strides="1,1"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1328" name="data_add_2418524190117321345" type="Const" version="opset1">
+ <data element_type="f16" offset="122378" shape="1,32,1,1" size="64"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1329" name="bottleneck3_8/inner/dw1/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_8/inner/dw1/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1330" name="bottleneck3_8/inner/dw1/fn/weights31020399561175" type="Const" version="opset1">
+ <data element_type="f32" offset="1576" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1331" name="bottleneck3_8/inner/dw1/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_8/inner/dw1/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1332" name="bottleneck3_8/dim_inc/conv/fq_input_0" type="FakeQuantize" version="opset1">
+ <data auto_broadcast="numpy" levels="256"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1"/>
+ <port id="2"/>
+ <port id="3"/>
+ <port id="4"/>
+ </input>
+ <output>
+ <port id="5" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1333" name="bottleneck3_8/dim_inc/bn/mean/Fused_Mul__copy117710281/quantized1194421528" type="Const" version="opset1">
+ <data element_type="i8" offset="122442" shape="128,32,1,1" size="4096"/>
+ <output>
+ <port id="0" precision="I8">
+ <dim>128</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1334" name="bottleneck3_8/dim_inc/bn/mean/Fused_Mul__copy117710281/quantized/to_f16" type="Convert" version="opset1">
+ <data destination_type="f16"/>
+ <input>
+ <port id="0">
+ <dim>128</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="1" precision="FP16">
+ <dim>128</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1335" name="bottleneck3_8/dim_inc/conv/fq_weights_1/zero_point1195720862" type="Const" version="opset1">
+ <data element_type="f16" offset="126538" shape="128,1,1,1" size="256"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1336" name="bottleneck3_8/dim_inc/conv/fq_weights_1/minus_zp" type="Subtract" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>128</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="1">
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>128</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1337" name="bottleneck3_8/dim_inc/conv/fq_weights_1/scale1195221756" type="Const" version="opset1">
+ <data element_type="f16" offset="126794" shape="128,1,1,1" size="256"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1338" name="bottleneck3_8/dim_inc/conv/fq_weights_1/mulpiply_by_scale" type="Multiply" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>128</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="1">
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>128</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1339" name="bottleneck3_8/dim_inc/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1,1" pads_begin="0,0" pads_end="0,0" strides="1,1"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1">
+ <dim>128</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1340" name="data_add_2419324198117920205" type="Const" version="opset1">
+ <data element_type="f16" offset="127050" shape="1,128,1,1" size="256"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1341" name="bottleneck3_8/dim_inc/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_8/dim_inc/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1342" name="bottleneck3_8/add/fq_input_1" type="FakeQuantize" version="opset1">
+ <data auto_broadcast="numpy" levels="256"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1"/>
+ <port id="2"/>
+ <port id="3"/>
+ <port id="4"/>
+ </input>
+ <output>
+ <port id="5" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1343" name="bottleneck3_8/add" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_8/add" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1344" name="bottleneck3_8/fn/weights31160404061182" type="Const" version="opset1">
+ <data element_type="f32" offset="1576" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1345" name="bottleneck3_8/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_8/add" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1346" name="bottleneck3_9/add/fq_input_0" type="FakeQuantize" version="opset1">
+ <data auto_broadcast="numpy" levels="256"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1"/>
+ <port id="2"/>
+ <port id="3"/>
+ <port id="4"/>
+ </input>
+ <output>
+ <port id="5" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1347" name="4514451822098" type="Const" version="opset1">
+ <data element_type="f16" offset="127306" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="1348" name="4515451919431" type="Const" version="opset1">
+ <data element_type="f16" offset="127308" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="1349" name="4516452020946" type="Const" version="opset1">
+ <data element_type="f16" offset="127306" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="1350" name="4517452122110" type="Const" version="opset1">
+ <data element_type="f16" offset="127308" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="1351" name="4144414821894" type="Const" version="opset1">
+ <data element_type="f16" offset="127310" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="1352" name="4145414919884" type="Const" version="opset1">
+ <data element_type="f16" offset="49164" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="1353" name="4146415022497" type="Const" version="opset1">
+ <data element_type="f16" offset="127310" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="1354" name="4147415121213" type="Const" version="opset1">
+ <data element_type="f16" offset="49164" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="1355" name="3164316820091" type="Const" version="opset1">
+ <data element_type="f16" offset="127312" shape="1,32,1,1" size="64"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1356" name="3165316920031" type="Const" version="opset1">
+ <data element_type="f16" offset="127376" shape="1,32,1,1" size="64"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1357" name="3166317021051" type="Const" version="opset1">
+ <data element_type="f16" offset="127312" shape="1,32,1,1" size="64"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1358" name="3167317120763" type="Const" version="opset1">
+ <data element_type="f16" offset="127376" shape="1,32,1,1" size="64"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1359" name="bottleneck3_9/dim_red/bn/mean/Fused_Mul__copy118410284/quantized1168019614" type="Const" version="opset1">
+ <data element_type="i8" offset="127440" shape="32,128,1,1" size="4096"/>
+ <output>
+ <port id="0" precision="I8">
+ <dim>32</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1360" name="bottleneck3_9/dim_red/bn/mean/Fused_Mul__copy118410284/quantized/to_f16" type="Convert" version="opset1">
+ <data destination_type="f16"/>
+ <input>
+ <port id="0">
+ <dim>32</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="1" precision="FP16">
+ <dim>32</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1361" name="bottleneck3_9/dim_red/conv/fq_weights_1/zero_point1169322359" type="Const" version="opset1">
+ <data element_type="f16" offset="131536" shape="32,1,1,1" size="64"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1362" name="bottleneck3_9/dim_red/conv/fq_weights_1/minus_zp" type="Subtract" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>32</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="1">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>32</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1363" name="bottleneck3_9/dim_red/conv/fq_weights_1/scale1168820730" type="Const" version="opset1">
+ <data element_type="f16" offset="131600" shape="32,1,1,1" size="64"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1364" name="bottleneck3_9/dim_red/conv/fq_weights_1/mulpiply_by_scale" type="Multiply" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>32</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="1">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>32</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1365" name="bottleneck3_9/dim_red/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1,1" pads_begin="0,0" pads_end="0,0" strides="1,1"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1">
+ <dim>32</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1366" name="data_add_2420124206118620067" type="Const" version="opset1">
+ <data element_type="f16" offset="131664" shape="1,32,1,1" size="64"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1367" name="bottleneck3_9/dim_red/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_9/dim_red/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1368" name="bottleneck3_9/dim_red/fn/weights30848398691188" type="Const" version="opset1">
+ <data element_type="f32" offset="1576" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1369" name="bottleneck3_9/dim_red/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_9/dim_red/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1370" name="bottleneck3_9/inner/dw1/conv/fq_input_0" type="FakeQuantize" version="opset1">
+ <data auto_broadcast="numpy" levels="256"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="2">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="3">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="4">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="5" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1371" name="16819/value1682120793" type="Const" version="opset1">
+ <data element_type="i64" offset="43778" shape="5" size="40"/>
+ <output>
+ <port id="0" precision="I64">
+ <dim>5</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1372" name="bottleneck3_9/inner/dw1/bn/mean/Fused_Mul__copy119010287/quantized1184819482" type="Const" version="opset1">
+ <data element_type="i8" offset="131728" shape="32,1,3,3" size="288"/>
+ <output>
+ <port id="0" precision="I8">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1373" name="bottleneck3_9/inner/dw1/bn/mean/Fused_Mul__copy119010287/quantized/to_f16" type="Convert" version="opset1">
+ <data destination_type="f16"/>
+ <input>
+ <port id="0">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </input>
+ <output>
+ <port id="1" precision="FP16">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1374" name="bottleneck3_9/inner/dw1/conv/fq_weights_1/zero_point1186120949" type="Const" version="opset1">
+ <data element_type="f16" offset="132016" shape="32,1,1,1" size="64"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1375" name="bottleneck3_9/inner/dw1/conv/fq_weights_1/minus_zp" type="Subtract" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ <port id="1">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1376" name="bottleneck3_9/inner/dw1/conv/fq_weights_1/scale1185621012" type="Const" version="opset1">
+ <data element_type="f16" offset="132080" shape="32,1,1,1" size="64"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1377" name="bottleneck3_9/inner/dw1/conv/fq_weights_1/mulpiply_by_scale" type="Multiply" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ <port id="1">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1378" name="16819" type="Reshape" version="opset1">
+ <data special_zero="true"/>
+ <input>
+ <port id="0">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ <port id="1">
+ <dim>5</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1379" name="bottleneck3_9/inner/dw1/conv" type="GroupConvolution" version="opset1">
+ <data auto_pad="explicit" dilations="1,1" pads_begin="1,1" pads_end="1,1" strides="1,1"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1380" name="data_add_2420924214119219692" type="Const" version="opset1">
+ <data element_type="f16" offset="132144" shape="1,32,1,1" size="64"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1381" name="bottleneck3_9/inner/dw1/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_9/inner/dw1/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1382" name="bottleneck3_9/inner/dw1/fn/weights30768404031194" type="Const" version="opset1">
+ <data element_type="f32" offset="1576" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1383" name="bottleneck3_9/inner/dw1/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_9/inner/dw1/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1384" name="bottleneck3_9/dim_inc/conv/fq_input_0" type="FakeQuantize" version="opset1">
+ <data auto_broadcast="numpy" levels="256"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1"/>
+ <port id="2"/>
+ <port id="3"/>
+ <port id="4"/>
+ </input>
+ <output>
+ <port id="5" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1385" name="bottleneck3_9/dim_inc/bn/mean/Fused_Mul__copy119610290/quantized1403221969" type="Const" version="opset1">
+ <data element_type="i8" offset="132208" shape="128,32,1,1" size="4096"/>
+ <output>
+ <port id="0" precision="I8">
+ <dim>128</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1386" name="bottleneck3_9/dim_inc/bn/mean/Fused_Mul__copy119610290/quantized/to_f16" type="Convert" version="opset1">
+ <data destination_type="f16"/>
+ <input>
+ <port id="0">
+ <dim>128</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="1" precision="FP16">
+ <dim>128</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1387" name="bottleneck3_9/dim_inc/conv/fq_weights_1/zero_point1404521216" type="Const" version="opset1">
+ <data element_type="f16" offset="136304" shape="128,1,1,1" size="256"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1388" name="bottleneck3_9/dim_inc/conv/fq_weights_1/minus_zp" type="Subtract" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>128</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="1">
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>128</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1389" name="bottleneck3_9/dim_inc/conv/fq_weights_1/scale1404021906" type="Const" version="opset1">
+ <data element_type="f16" offset="136560" shape="128,1,1,1" size="256"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1390" name="bottleneck3_9/dim_inc/conv/fq_weights_1/mulpiply_by_scale" type="Multiply" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>128</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="1">
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>128</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1391" name="bottleneck3_9/dim_inc/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1,1" pads_begin="0,0" pads_end="0,0" strides="1,1"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1">
+ <dim>128</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1392" name="data_add_2421724222119821699" type="Const" version="opset1">
+ <data element_type="f16" offset="136816" shape="1,128,1,1" size="256"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1393" name="bottleneck3_9/dim_inc/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_9/dim_inc/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1394" name="bottleneck3_9/add/fq_input_1" type="FakeQuantize" version="opset1">
+ <data auto_broadcast="numpy" levels="256"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1"/>
+ <port id="2"/>
+ <port id="3"/>
+ <port id="4"/>
+ </input>
+ <output>
+ <port id="5" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1395" name="bottleneck3_9/add" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_9/add" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1396" name="bottleneck3_9/fn/weights31024406761201" type="Const" version="opset1">
+ <data element_type="f32" offset="1576" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1397" name="bottleneck3_9/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_9/add" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1398" name="bottleneck3_10/add/fq_input_0" type="FakeQuantize" version="opset1">
+ <data auto_broadcast="numpy" levels="256"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1"/>
+ <port id="2"/>
+ <port id="3"/>
+ <port id="4"/>
+ </input>
+ <output>
+ <port id="5" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1399" name="4474447819539" type="Const" version="opset1">
+ <data element_type="f16" offset="137072" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="1400" name="4475447919491" type="Const" version="opset1">
+ <data element_type="f16" offset="137074" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="1401" name="4476448020022" type="Const" version="opset1">
+ <data element_type="f16" offset="137072" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="1402" name="4477448121588" type="Const" version="opset1">
+ <data element_type="f16" offset="137074" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="1403" name="2944294820112" type="Const" version="opset1">
+ <data element_type="f16" offset="137076" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="1404" name="2945294919929" type="Const" version="opset1">
+ <data element_type="f16" offset="137078" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="1405" name="2946295019737" type="Const" version="opset1">
+ <data element_type="f16" offset="137076" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="1406" name="2947295120466" type="Const" version="opset1">
+ <data element_type="f16" offset="137078" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="1407" name="4224422820349" type="Const" version="opset1">
+ <data element_type="f16" offset="137080" shape="1,32,1,1" size="64"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1408" name="4225422922002" type="Const" version="opset1">
+ <data element_type="f16" offset="137144" shape="1,32,1,1" size="64"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1409" name="4226423020208" type="Const" version="opset1">
+ <data element_type="f16" offset="137080" shape="1,32,1,1" size="64"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1410" name="4227423122239" type="Const" version="opset1">
+ <data element_type="f16" offset="137144" shape="1,32,1,1" size="64"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1411" name="bottleneck3_10/dim_red/bn/mean/Fused_Mul__copy120310293/quantized1235221372" type="Const" version="opset1">
+ <data element_type="i8" offset="137208" shape="32,128,1,1" size="4096"/>
+ <output>
+ <port id="0" precision="I8">
+ <dim>32</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1412" name="bottleneck3_10/dim_red/bn/mean/Fused_Mul__copy120310293/quantized/to_f16" type="Convert" version="opset1">
+ <data destination_type="f16"/>
+ <input>
+ <port id="0">
+ <dim>32</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="1" precision="FP16">
+ <dim>32</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1413" name="bottleneck3_10/dim_red/conv/fq_weights_1/zero_point1236520145" type="Const" version="opset1">
+ <data element_type="f16" offset="141304" shape="32,1,1,1" size="64"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1414" name="bottleneck3_10/dim_red/conv/fq_weights_1/minus_zp" type="Subtract" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>32</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="1">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>32</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1415" name="bottleneck3_10/dim_red/conv/fq_weights_1/scale1236020934" type="Const" version="opset1">
+ <data element_type="f16" offset="141368" shape="32,1,1,1" size="64"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1416" name="bottleneck3_10/dim_red/conv/fq_weights_1/mulpiply_by_scale" type="Multiply" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>32</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="1">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>32</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1417" name="bottleneck3_10/dim_red/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1,1" pads_begin="0,0" pads_end="0,0" strides="1,1"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1">
+ <dim>32</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1418" name="data_add_2422524230120519773" type="Const" version="opset1">
+ <data element_type="f16" offset="141432" shape="1,32,1,1" size="64"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1419" name="bottleneck3_10/dim_red/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_10/dim_red/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1420" name="bottleneck3_10/dim_red/fn/weights30976405021207" type="Const" version="opset1">
+ <data element_type="f32" offset="1576" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1421" name="bottleneck3_10/dim_red/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_10/dim_red/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1422" name="bottleneck3_10/inner/dw1/conv/fq_input_0" type="FakeQuantize" version="opset1">
+ <data auto_broadcast="numpy" levels="256"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="2">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="3">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="4">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="5" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1423" name="16875/value1687722320" type="Const" version="opset1">
+ <data element_type="i64" offset="43778" shape="5" size="40"/>
+ <output>
+ <port id="0" precision="I64">
+ <dim>5</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1424" name="bottleneck3_10/inner/dw1/bn/mean/Fused_Mul__copy120910296/quantized1369620472" type="Const" version="opset1">
+ <data element_type="i8" offset="141496" shape="32,1,3,3" size="288"/>
+ <output>
+ <port id="0" precision="I8">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1425" name="bottleneck3_10/inner/dw1/bn/mean/Fused_Mul__copy120910296/quantized/to_f16" type="Convert" version="opset1">
+ <data destination_type="f16"/>
+ <input>
+ <port id="0">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </input>
+ <output>
+ <port id="1" precision="FP16">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1426" name="bottleneck3_10/inner/dw1/conv/fq_weights_1/zero_point1370919593" type="Const" version="opset1">
+ <data element_type="f16" offset="141784" shape="32,1,1,1" size="64"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1427" name="bottleneck3_10/inner/dw1/conv/fq_weights_1/minus_zp" type="Subtract" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ <port id="1">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1428" name="bottleneck3_10/inner/dw1/conv/fq_weights_1/scale1370420757" type="Const" version="opset1">
+ <data element_type="f16" offset="141848" shape="32,1,1,1" size="64"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1429" name="bottleneck3_10/inner/dw1/conv/fq_weights_1/mulpiply_by_scale" type="Multiply" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ <port id="1">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1430" name="16875" type="Reshape" version="opset1">
+ <data special_zero="true"/>
+ <input>
+ <port id="0">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ <port id="1">
+ <dim>5</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1431" name="bottleneck3_10/inner/dw1/conv" type="GroupConvolution" version="opset1">
+ <data auto_pad="explicit" dilations="1,1" pads_begin="1,1" pads_end="1,1" strides="1,1"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1432" name="data_add_2423324238121120427" type="Const" version="opset1">
+ <data element_type="f16" offset="141912" shape="1,32,1,1" size="64"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1433" name="bottleneck3_10/inner/dw1/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_10/inner/dw1/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1434" name="bottleneck3_10/inner/dw1/fn/weights31016401511213" type="Const" version="opset1">
+ <data element_type="f32" offset="1576" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1435" name="bottleneck3_10/inner/dw1/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_10/inner/dw1/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1436" name="bottleneck3_10/dim_inc/conv/fq_input_0" type="FakeQuantize" version="opset1">
+ <data auto_broadcast="numpy" levels="256"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1"/>
+ <port id="2"/>
+ <port id="3"/>
+ <port id="4"/>
+ </input>
+ <output>
+ <port id="5" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1437" name="bottleneck3_10/dim_inc/bn/mean/Fused_Mul__copy121510299/quantized1151222302" type="Const" version="opset1">
+ <data element_type="i8" offset="141976" shape="128,32,1,1" size="4096"/>
+ <output>
+ <port id="0" precision="I8">
+ <dim>128</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1438" name="bottleneck3_10/dim_inc/bn/mean/Fused_Mul__copy121510299/quantized/to_f16" type="Convert" version="opset1">
+ <data destination_type="f16"/>
+ <input>
+ <port id="0">
+ <dim>128</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="1" precision="FP16">
+ <dim>128</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1439" name="bottleneck3_10/dim_inc/conv/fq_weights_1/zero_point1152520049" type="Const" version="opset1">
+ <data element_type="f16" offset="146072" shape="128,1,1,1" size="256"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1440" name="bottleneck3_10/dim_inc/conv/fq_weights_1/minus_zp" type="Subtract" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>128</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="1">
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>128</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1441" name="bottleneck3_10/dim_inc/conv/fq_weights_1/scale1152021642" type="Const" version="opset1">
+ <data element_type="f16" offset="146328" shape="128,1,1,1" size="256"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1442" name="bottleneck3_10/dim_inc/conv/fq_weights_1/mulpiply_by_scale" type="Multiply" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>128</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="1">
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>128</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1443" name="bottleneck3_10/dim_inc/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1,1" pads_begin="0,0" pads_end="0,0" strides="1,1"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1">
+ <dim>128</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1444" name="data_add_2424124246121722014" type="Const" version="opset1">
+ <data element_type="f16" offset="146584" shape="1,128,1,1" size="256"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1445" name="bottleneck3_10/dim_inc/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_10/dim_inc/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1446" name="bottleneck3_10/add/fq_input_1" type="FakeQuantize" version="opset1">
+ <data auto_broadcast="numpy" levels="256"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1"/>
+ <port id="2"/>
+ <port id="3"/>
+ <port id="4"/>
+ </input>
+ <output>
+ <port id="5" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1447" name="bottleneck3_10/add" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_10/add" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1448" name="bottleneck3_10/fn/weights30940401451220" type="Const" version="opset1">
+ <data element_type="f32" offset="1576" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1449" name="bottleneck3_10/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_10/add" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1450" name="bottleneck4_0/dim_red/conv/fq_input_0" type="FakeQuantize" version="opset1">
+ <data auto_broadcast="numpy" levels="256"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1"/>
+ <port id="2"/>
+ <port id="3"/>
+ <port id="4"/>
+ </input>
+ <output>
+ <port id="5" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1451" name="bottleneck4_0/skip/pooling" type="MaxPool" version="opset1">
+ <data auto_pad="explicit" kernel="2,2" pads_begin="0,0" pads_end="0,0" rounding_type="ceil" strides="2,2"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </input>
+ <output>
+ <port id="1" names="bottleneck4_0/skip/pooling" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1452" name="bottleneck4_0/skip/bn/mean/Fused_Mul__copy122310302/quantized1266421063" type="Const" version="opset1">
+ <data element_type="i8" offset="146840" shape="256,128,1,1" size="32768"/>
+ <output>
+ <port id="0" precision="I8">
+ <dim>256</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1453" name="bottleneck4_0/skip/bn/mean/Fused_Mul__copy122310302/quantized/to_f16" type="Convert" version="opset1">
+ <data destination_type="f16"/>
+ <input>
+ <port id="0">
+ <dim>256</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="1" precision="FP16">
+ <dim>256</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1454" name="bottleneck4_0/skip/conv/fq_weights_1/zero_point1267719407" type="Const" version="opset1">
+ <data element_type="f16" offset="179608" shape="256,1,1,1" size="512"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1455" name="bottleneck4_0/skip/conv/fq_weights_1/minus_zp" type="Subtract" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>256</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="1">
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>256</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1456" name="bottleneck4_0/skip/conv/fq_weights_1/scale1267219896" type="Const" version="opset1">
+ <data element_type="f16" offset="180120" shape="256,1,1,1" size="512"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1457" name="bottleneck4_0/skip/conv/fq_weights_1/mulpiply_by_scale" type="Multiply" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>256</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="1">
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>256</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1458" name="bottleneck4_0/skip/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1,1" pads_begin="0,0" pads_end="0,0" strides="1,1"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1">
+ <dim>256</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1459" name="data_add_2424924254122520880" type="Const" version="opset1">
+ <data element_type="f16" offset="180632" shape="1,256,1,1" size="512"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1460" name="bottleneck4_0/skip/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_0/skip/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1461" name="bottleneck4_0/add/fq_input_0" type="FakeQuantize" version="opset1">
+ <data auto_broadcast="numpy" levels="256"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1"/>
+ <port id="2"/>
+ <port id="3"/>
+ <port id="4"/>
+ </input>
+ <output>
+ <port id="5" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1462" name="3114311822758" type="Const" version="opset1">
+ <data element_type="f16" offset="181144" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="1463" name="3115311920622" type="Const" version="opset1">
+ <data element_type="f16" offset="181146" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="1464" name="3116312020928" type="Const" version="opset1">
+ <data element_type="f16" offset="181144" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="1465" name="3117312121978" type="Const" version="opset1">
+ <data element_type="f16" offset="181146" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="1466" name="3004300822833" type="Const" version="opset1">
+ <data element_type="f16" offset="181148" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="1467" name="3005300922149" type="Const" version="opset1">
+ <data element_type="f16" offset="181150" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="1468" name="3006301022056" type="Const" version="opset1">
+ <data element_type="f16" offset="181148" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="1469" name="3007301120124" type="Const" version="opset1">
+ <data element_type="f16" offset="181150" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="1470" name="4244424822983" type="Const" version="opset1">
+ <data element_type="f16" offset="181152" shape="1,64,1,1" size="128"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1471" name="4245424921591" type="Const" version="opset1">
+ <data element_type="f16" offset="181280" shape="1,64,1,1" size="128"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1472" name="4246425019413" type="Const" version="opset1">
+ <data element_type="f16" offset="181152" shape="1,64,1,1" size="128"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1473" name="4247425121324" type="Const" version="opset1">
+ <data element_type="f16" offset="181280" shape="1,64,1,1" size="128"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1474" name="bottleneck4_0/dim_red/bn/mean/Fused_Mul__copy122710304/quantized1182421375" type="Const" version="opset1">
+ <data element_type="i8" offset="181408" shape="64,128,1,1" size="8192"/>
+ <output>
+ <port id="0" precision="I8">
+ <dim>64</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1475" name="bottleneck4_0/dim_red/bn/mean/Fused_Mul__copy122710304/quantized/to_f16" type="Convert" version="opset1">
+ <data destination_type="f16"/>
+ <input>
+ <port id="0">
+ <dim>64</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="1" precision="FP16">
+ <dim>64</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1476" name="bottleneck4_0/dim_red/conv/fq_weights_1/zero_point1183720238" type="Const" version="opset1">
+ <data element_type="f16" offset="189600" shape="64,1,1,1" size="128"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1477" name="bottleneck4_0/dim_red/conv/fq_weights_1/minus_zp" type="Subtract" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>64</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="1">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>64</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1478" name="bottleneck4_0/dim_red/conv/fq_weights_1/scale1183221075" type="Const" version="opset1">
+ <data element_type="f16" offset="189728" shape="64,1,1,1" size="128"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1479" name="bottleneck4_0/dim_red/conv/fq_weights_1/mulpiply_by_scale" type="Multiply" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>64</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="1">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>64</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1480" name="bottleneck4_0/dim_red/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1,1" pads_begin="0,0" pads_end="0,0" strides="1,1"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1">
+ <dim>64</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1481" name="data_add_2425724262122922353" type="Const" version="opset1">
+ <data element_type="f16" offset="189856" shape="1,64,1,1" size="128"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1482" name="bottleneck4_0/dim_red/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_0/dim_red/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1483" name="bottleneck4_0/dim_red/fn/weights30984402531231" type="Const" version="opset1">
+ <data element_type="f32" offset="1576" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1484" name="bottleneck4_0/dim_red/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_0/dim_red/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1485" name="bottleneck4_0/inner/dw1/conv/fq_input_0" type="FakeQuantize" version="opset1">
+ <data auto_broadcast="numpy" levels="256"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="2">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="3">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="4">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="5" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1486" name="16879/value1688121138" type="Const" version="opset1">
+ <data element_type="i64" offset="189984" shape="5" size="40"/>
+ <output>
+ <port id="0" precision="I64">
+ <dim>5</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1487" name="bottleneck4_0/inner/dw1/bn/mean/Fused_Mul__copy123310307/quantized1393620607" type="Const" version="opset1">
+ <data element_type="i8" offset="190024" shape="64,1,3,3" size="576"/>
+ <output>
+ <port id="0" precision="I8">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1488" name="bottleneck4_0/inner/dw1/bn/mean/Fused_Mul__copy123310307/quantized/to_f16" type="Convert" version="opset1">
+ <data destination_type="f16"/>
+ <input>
+ <port id="0">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </input>
+ <output>
+ <port id="1" precision="FP16">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1489" name="bottleneck4_0/inner/dw1/conv/fq_weights_1/zero_point1394920988" type="Const" version="opset1">
+ <data element_type="f16" offset="190600" shape="64,1,1,1" size="128"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1490" name="bottleneck4_0/inner/dw1/conv/fq_weights_1/minus_zp" type="Subtract" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ <port id="1">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1491" name="bottleneck4_0/inner/dw1/conv/fq_weights_1/scale1394419923" type="Const" version="opset1">
+ <data element_type="f16" offset="190728" shape="64,1,1,1" size="128"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1492" name="bottleneck4_0/inner/dw1/conv/fq_weights_1/mulpiply_by_scale" type="Multiply" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ <port id="1">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1493" name="16879" type="Reshape" version="opset1">
+ <data special_zero="true"/>
+ <input>
+ <port id="0">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ <port id="1">
+ <dim>5</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1494" name="bottleneck4_0/inner/dw1/conv" type="GroupConvolution" version="opset1">
+ <data auto_pad="explicit" dilations="1,1" pads_begin="1,1" pads_end="1,1" strides="2,2"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1495" name="data_add_2426524270123522653" type="Const" version="opset1">
+ <data element_type="f16" offset="190856" shape="1,64,1,1" size="128"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1496" name="bottleneck4_0/inner/dw1/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_0/inner/dw1/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1497" name="bottleneck4_0/inner/dw1/fn/weights31072399351237" type="Const" version="opset1">
+ <data element_type="f32" offset="1576" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1498" name="bottleneck4_0/inner/dw1/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_0/inner/dw1/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1499" name="bottleneck4_0/dim_inc/conv/fq_input_0" type="FakeQuantize" version="opset1">
+ <data auto_broadcast="numpy" levels="256"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1"/>
+ <port id="2"/>
+ <port id="3"/>
+ <port id="4"/>
+ </input>
+ <output>
+ <port id="5" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1500" name="bottleneck4_0/dim_inc/bn/mean/Fused_Mul__copy123910310/quantized1326420133" type="Const" version="opset1">
+ <data element_type="i8" offset="190984" shape="256,64,1,1" size="16384"/>
+ <output>
+ <port id="0" precision="I8">
+ <dim>256</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1501" name="bottleneck4_0/dim_inc/bn/mean/Fused_Mul__copy123910310/quantized/to_f16" type="Convert" version="opset1">
+ <data destination_type="f16"/>
+ <input>
+ <port id="0">
+ <dim>256</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="1" precision="FP16">
+ <dim>256</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1502" name="bottleneck4_0/dim_inc/conv/fq_weights_1/zero_point1327721255" type="Const" version="opset1">
+ <data element_type="f16" offset="207368" shape="256,1,1,1" size="512"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1503" name="bottleneck4_0/dim_inc/conv/fq_weights_1/minus_zp" type="Subtract" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>256</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="1">
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>256</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1504" name="bottleneck4_0/dim_inc/conv/fq_weights_1/scale1327221849" type="Const" version="opset1">
+ <data element_type="f16" offset="207880" shape="256,1,1,1" size="512"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1505" name="bottleneck4_0/dim_inc/conv/fq_weights_1/mulpiply_by_scale" type="Multiply" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>256</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="1">
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>256</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1506" name="bottleneck4_0/dim_inc/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1,1" pads_begin="0,0" pads_end="0,0" strides="1,1"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1">
+ <dim>256</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1507" name="data_add_2427324278124122158" type="Const" version="opset1">
+ <data element_type="f16" offset="208392" shape="1,256,1,1" size="512"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1508" name="bottleneck4_0/dim_inc/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_0/dim_inc/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1509" name="bottleneck4_0/add/fq_input_1" type="FakeQuantize" version="opset1">
+ <data auto_broadcast="numpy" levels="256"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1"/>
+ <port id="2"/>
+ <port id="3"/>
+ <port id="4"/>
+ </input>
+ <output>
+ <port id="5" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1510" name="bottleneck4_0/add" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_0/add" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1511" name="bottleneck4_0/fn/weights31080401151244" type="Const" version="opset1">
+ <data element_type="f32" offset="1576" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1512" name="bottleneck4_0/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_0/add" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1513" name="bottleneck4_1/add/fq_input_0" type="FakeQuantize" version="opset1">
+ <data auto_broadcast="numpy" levels="256"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1"/>
+ <port id="2"/>
+ <port id="3"/>
+ <port id="4"/>
+ </input>
+ <output>
+ <port id="5" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1514" name="3974397819956" type="Const" version="opset1">
+ <data element_type="f16" offset="208904" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="1515" name="3975397919617" type="Const" version="opset1">
+ <data element_type="f16" offset="208906" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="1516" name="3976398019632" type="Const" version="opset1">
+ <data element_type="f16" offset="208904" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="1517" name="3977398122686" type="Const" version="opset1">
+ <data element_type="f16" offset="208906" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="1518" name="4544454822989" type="Const" version="opset1">
+ <data element_type="f16" offset="208908" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="1519" name="4545454920721" type="Const" version="opset1">
+ <data element_type="f16" offset="208910" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="1520" name="4546455021402" type="Const" version="opset1">
+ <data element_type="f16" offset="208908" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="1521" name="4547455119971" type="Const" version="opset1">
+ <data element_type="f16" offset="208910" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="1522" name="5104510821027" type="Const" version="opset1">
+ <data element_type="f16" offset="208912" shape="1,64,1,1" size="128"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1523" name="5105510922038" type="Const" version="opset1">
+ <data element_type="f16" offset="209040" shape="1,64,1,1" size="128"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1524" name="5106511021408" type="Const" version="opset1">
+ <data element_type="f16" offset="208912" shape="1,64,1,1" size="128"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1525" name="5107511121114" type="Const" version="opset1">
+ <data element_type="f16" offset="209040" shape="1,64,1,1" size="128"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1526" name="bottleneck4_1/dim_red/bn/mean/Fused_Mul__copy124610313/quantized1180020589" type="Const" version="opset1">
+ <data element_type="i8" offset="209168" shape="64,256,1,1" size="16384"/>
+ <output>
+ <port id="0" precision="I8">
+ <dim>64</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1527" name="bottleneck4_1/dim_red/bn/mean/Fused_Mul__copy124610313/quantized/to_f16" type="Convert" version="opset1">
+ <data destination_type="f16"/>
+ <input>
+ <port id="0">
+ <dim>64</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="1" precision="FP16">
+ <dim>64</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1528" name="bottleneck4_1/dim_red/conv/fq_weights_1/zero_point1181319629" type="Const" version="opset1">
+ <data element_type="f16" offset="225552" shape="64,1,1,1" size="128"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1529" name="bottleneck4_1/dim_red/conv/fq_weights_1/minus_zp" type="Subtract" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>64</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="1">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>64</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1530" name="bottleneck4_1/dim_red/conv/fq_weights_1/scale1180820541" type="Const" version="opset1">
+ <data element_type="f16" offset="225680" shape="64,1,1,1" size="128"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1531" name="bottleneck4_1/dim_red/conv/fq_weights_1/mulpiply_by_scale" type="Multiply" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>64</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="1">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>64</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1532" name="bottleneck4_1/dim_red/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1,1" pads_begin="0,0" pads_end="0,0" strides="1,1"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1">
+ <dim>64</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1533" name="data_add_2428124286124820064" type="Const" version="opset1">
+ <data element_type="f16" offset="225808" shape="1,64,1,1" size="128"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1534" name="bottleneck4_1/dim_red/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_1/dim_red/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1535" name="bottleneck4_1/dim_red/fn/weights31040401781250" type="Const" version="opset1">
+ <data element_type="f32" offset="1576" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1536" name="bottleneck4_1/dim_red/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_1/dim_red/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1537" name="bottleneck4_1/inner/dw1/conv/fq_input_0" type="FakeQuantize" version="opset1">
+ <data auto_broadcast="numpy" levels="256"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="2">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="3">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="4">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="5" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1538" name="16923/value1692520649" type="Const" version="opset1">
+ <data element_type="i64" offset="189984" shape="5" size="40"/>
+ <output>
+ <port id="0" precision="I64">
+ <dim>5</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1539" name="bottleneck4_1/inner/dw1/bn/mean/Fused_Mul__copy125210316/quantized1379221036" type="Const" version="opset1">
+ <data element_type="i8" offset="225936" shape="64,1,3,3" size="576"/>
+ <output>
+ <port id="0" precision="I8">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1540" name="bottleneck4_1/inner/dw1/bn/mean/Fused_Mul__copy125210316/quantized/to_f16" type="Convert" version="opset1">
+ <data destination_type="f16"/>
+ <input>
+ <port id="0">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </input>
+ <output>
+ <port id="1" precision="FP16">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1541" name="bottleneck4_1/inner/dw1/conv/fq_weights_1/zero_point1380522875" type="Const" version="opset1">
+ <data element_type="f16" offset="226512" shape="64,1,1,1" size="128"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1542" name="bottleneck4_1/inner/dw1/conv/fq_weights_1/minus_zp" type="Subtract" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ <port id="1">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1543" name="bottleneck4_1/inner/dw1/conv/fq_weights_1/scale1380021909" type="Const" version="opset1">
+ <data element_type="f16" offset="226640" shape="64,1,1,1" size="128"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1544" name="bottleneck4_1/inner/dw1/conv/fq_weights_1/mulpiply_by_scale" type="Multiply" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ <port id="1">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1545" name="16923" type="Reshape" version="opset1">
+ <data special_zero="true"/>
+ <input>
+ <port id="0">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ <port id="1">
+ <dim>5</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1546" name="bottleneck4_1/inner/dw1/conv" type="GroupConvolution" version="opset1">
+ <data auto_pad="explicit" dilations="1,1" pads_begin="1,1" pads_end="1,1" strides="1,1"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1547" name="data_add_2428924294125422251" type="Const" version="opset1">
+ <data element_type="f16" offset="226768" shape="1,64,1,1" size="128"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1548" name="bottleneck4_1/inner/dw1/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_1/inner/dw1/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1549" name="bottleneck4_1/inner/dw1/fn/weights30964399921256" type="Const" version="opset1">
+ <data element_type="f32" offset="1576" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1550" name="bottleneck4_1/inner/dw1/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_1/inner/dw1/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1551" name="bottleneck4_1/dim_inc/conv/fq_input_0" type="FakeQuantize" version="opset1">
+ <data auto_broadcast="numpy" levels="256"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1"/>
+ <port id="2"/>
+ <port id="3"/>
+ <port id="4"/>
+ </input>
+ <output>
+ <port id="5" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1552" name="bottleneck4_1/dim_inc/bn/mean/Fused_Mul__copy125810319/quantized1249620772" type="Const" version="opset1">
+ <data element_type="i8" offset="226896" shape="256,64,1,1" size="16384"/>
+ <output>
+ <port id="0" precision="I8">
+ <dim>256</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1553" name="bottleneck4_1/dim_inc/bn/mean/Fused_Mul__copy125810319/quantized/to_f16" type="Convert" version="opset1">
+ <data destination_type="f16"/>
+ <input>
+ <port id="0">
+ <dim>256</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="1" precision="FP16">
+ <dim>256</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1554" name="bottleneck4_1/dim_inc/conv/fq_weights_1/zero_point1250921828" type="Const" version="opset1">
+ <data element_type="f16" offset="243280" shape="256,1,1,1" size="512"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1555" name="bottleneck4_1/dim_inc/conv/fq_weights_1/minus_zp" type="Subtract" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>256</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="1">
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>256</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1556" name="bottleneck4_1/dim_inc/conv/fq_weights_1/scale1250422458" type="Const" version="opset1">
+ <data element_type="f16" offset="243792" shape="256,1,1,1" size="512"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1557" name="bottleneck4_1/dim_inc/conv/fq_weights_1/mulpiply_by_scale" type="Multiply" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>256</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="1">
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>256</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1558" name="bottleneck4_1/dim_inc/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1,1" pads_begin="0,0" pads_end="0,0" strides="1,1"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1">
+ <dim>256</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1559" name="data_add_2429724302126020523" type="Const" version="opset1">
+ <data element_type="f16" offset="244304" shape="1,256,1,1" size="512"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1560" name="bottleneck4_1/dim_inc/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_1/dim_inc/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1561" name="bottleneck4_1/add/fq_input_1" type="FakeQuantize" version="opset1">
+ <data auto_broadcast="numpy" levels="256"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1"/>
+ <port id="2"/>
+ <port id="3"/>
+ <port id="4"/>
+ </input>
+ <output>
+ <port id="5" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1562" name="bottleneck4_1/add" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_1/add" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1563" name="bottleneck4_1/fn/weights31132403611263" type="Const" version="opset1">
+ <data element_type="f32" offset="1576" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1564" name="bottleneck4_1/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_1/add" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1565" name="bottleneck4_2/add/fq_input_0" type="FakeQuantize" version="opset1">
+ <data auto_broadcast="numpy" levels="256"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1"/>
+ <port id="2"/>
+ <port id="3"/>
+ <port id="4"/>
+ </input>
+ <output>
+ <port id="5" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1566" name="2694269819551" type="Const" version="opset1">
+ <data element_type="f16" offset="244816" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="1567" name="2695269921813" type="Const" version="opset1">
+ <data element_type="f16" offset="244818" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="1568" name="2696270020040" type="Const" version="opset1">
+ <data element_type="f16" offset="244816" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="1569" name="2697270122179" type="Const" version="opset1">
+ <data element_type="f16" offset="244818" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="1570" name="4644464821657" type="Const" version="opset1">
+ <data element_type="f16" offset="244820" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="1571" name="4645464920733" type="Const" version="opset1">
+ <data element_type="f16" offset="244822" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="1572" name="4646465022779" type="Const" version="opset1">
+ <data element_type="f16" offset="244820" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="1573" name="4647465121354" type="Const" version="opset1">
+ <data element_type="f16" offset="244822" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="1574" name="4044404822188" type="Const" version="opset1">
+ <data element_type="f16" offset="244824" shape="1,64,1,1" size="128"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1575" name="4045404920319" type="Const" version="opset1">
+ <data element_type="f16" offset="244952" shape="1,64,1,1" size="128"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1576" name="4046405019905" type="Const" version="opset1">
+ <data element_type="f16" offset="244824" shape="1,64,1,1" size="128"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1577" name="4047405122782" type="Const" version="opset1">
+ <data element_type="f16" offset="244952" shape="1,64,1,1" size="128"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1578" name="bottleneck4_2/dim_red/bn/mean/Fused_Mul__copy126510322/quantized1242421189" type="Const" version="opset1">
+ <data element_type="i8" offset="245080" shape="64,256,1,1" size="16384"/>
+ <output>
+ <port id="0" precision="I8">
+ <dim>64</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1579" name="bottleneck4_2/dim_red/bn/mean/Fused_Mul__copy126510322/quantized/to_f16" type="Convert" version="opset1">
+ <data destination_type="f16"/>
+ <input>
+ <port id="0">
+ <dim>64</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="1" precision="FP16">
+ <dim>64</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1580" name="bottleneck4_2/dim_red/conv/fq_weights_1/zero_point1243720139" type="Const" version="opset1">
+ <data element_type="f16" offset="261464" shape="64,1,1,1" size="128"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1581" name="bottleneck4_2/dim_red/conv/fq_weights_1/minus_zp" type="Subtract" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>64</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="1">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>64</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1582" name="bottleneck4_2/dim_red/conv/fq_weights_1/scale1243221885" type="Const" version="opset1">
+ <data element_type="f16" offset="261592" shape="64,1,1,1" size="128"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1583" name="bottleneck4_2/dim_red/conv/fq_weights_1/mulpiply_by_scale" type="Multiply" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>64</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="1">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>64</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1584" name="bottleneck4_2/dim_red/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1,1" pads_begin="0,0" pads_end="0,0" strides="1,1"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1">
+ <dim>64</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1585" name="data_add_2430524310126719380" type="Const" version="opset1">
+ <data element_type="f16" offset="261720" shape="1,64,1,1" size="128"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1586" name="bottleneck4_2/dim_red/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_2/dim_red/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1587" name="bottleneck4_2/dim_red/fn/weights30812398181269" type="Const" version="opset1">
+ <data element_type="f32" offset="1576" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1588" name="bottleneck4_2/dim_red/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_2/dim_red/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1589" name="bottleneck4_2/inner/dw1/conv/fq_input_0" type="FakeQuantize" version="opset1">
+ <data auto_broadcast="numpy" levels="256"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="2">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="3">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="4">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="5" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1590" name="16867/value1686920148" type="Const" version="opset1">
+ <data element_type="i64" offset="189984" shape="5" size="40"/>
+ <output>
+ <port id="0" precision="I64">
+ <dim>5</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1591" name="bottleneck4_2/inner/dw1/bn/mean/Fused_Mul__copy127110325/quantized1175222380" type="Const" version="opset1">
+ <data element_type="i8" offset="261848" shape="64,1,3,3" size="576"/>
+ <output>
+ <port id="0" precision="I8">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1592" name="bottleneck4_2/inner/dw1/bn/mean/Fused_Mul__copy127110325/quantized/to_f16" type="Convert" version="opset1">
+ <data destination_type="f16"/>
+ <input>
+ <port id="0">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </input>
+ <output>
+ <port id="1" precision="FP16">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1593" name="bottleneck4_2/inner/dw1/conv/fq_weights_1/zero_point1176521537" type="Const" version="opset1">
+ <data element_type="f16" offset="262424" shape="64,1,1,1" size="128"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1594" name="bottleneck4_2/inner/dw1/conv/fq_weights_1/minus_zp" type="Subtract" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ <port id="1">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1595" name="bottleneck4_2/inner/dw1/conv/fq_weights_1/scale1176022944" type="Const" version="opset1">
+ <data element_type="f16" offset="262552" shape="64,1,1,1" size="128"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1596" name="bottleneck4_2/inner/dw1/conv/fq_weights_1/mulpiply_by_scale" type="Multiply" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ <port id="1">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1597" name="16867" type="Reshape" version="opset1">
+ <data special_zero="true"/>
+ <input>
+ <port id="0">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ <port id="1">
+ <dim>5</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1598" name="bottleneck4_2/inner/dw1/conv" type="GroupConvolution" version="opset1">
+ <data auto_pad="explicit" dilations="1,1" pads_begin="1,1" pads_end="1,1" strides="1,1"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1599" name="data_add_2431324318127320151" type="Const" version="opset1">
+ <data element_type="f16" offset="262680" shape="1,64,1,1" size="128"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1600" name="bottleneck4_2/inner/dw1/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_2/inner/dw1/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1601" name="bottleneck4_2/inner/dw1/fn/weights30960401241275" type="Const" version="opset1">
+ <data element_type="f32" offset="1576" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1602" name="bottleneck4_2/inner/dw1/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_2/inner/dw1/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1603" name="bottleneck4_2/dim_inc/conv/fq_input_0" type="FakeQuantize" version="opset1">
+ <data auto_broadcast="numpy" levels="256"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1"/>
+ <port id="2"/>
+ <port id="3"/>
+ <port id="4"/>
+ </input>
+ <output>
+ <port id="5" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1604" name="bottleneck4_2/dim_inc/bn/mean/Fused_Mul__copy127710328/quantized1343219902" type="Const" version="opset1">
+ <data element_type="i8" offset="262808" shape="256,64,1,1" size="16384"/>
+ <output>
+ <port id="0" precision="I8">
+ <dim>256</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1605" name="bottleneck4_2/dim_inc/bn/mean/Fused_Mul__copy127710328/quantized/to_f16" type="Convert" version="opset1">
+ <data destination_type="f16"/>
+ <input>
+ <port id="0">
+ <dim>256</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="1" precision="FP16">
+ <dim>256</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1606" name="bottleneck4_2/dim_inc/conv/fq_weights_1/zero_point1344521468" type="Const" version="opset1">
+ <data element_type="f16" offset="279192" shape="256,1,1,1" size="512"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1607" name="bottleneck4_2/dim_inc/conv/fq_weights_1/minus_zp" type="Subtract" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>256</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="1">
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>256</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1608" name="bottleneck4_2/dim_inc/conv/fq_weights_1/scale1344020652" type="Const" version="opset1">
+ <data element_type="f16" offset="279704" shape="256,1,1,1" size="512"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1609" name="bottleneck4_2/dim_inc/conv/fq_weights_1/mulpiply_by_scale" type="Multiply" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>256</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="1">
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>256</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1610" name="bottleneck4_2/dim_inc/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1,1" pads_begin="0,0" pads_end="0,0" strides="1,1"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1">
+ <dim>256</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1611" name="data_add_2432124326127920925" type="Const" version="opset1">
+ <data element_type="f16" offset="280216" shape="1,256,1,1" size="512"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1612" name="bottleneck4_2/dim_inc/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_2/dim_inc/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1613" name="bottleneck4_2/add/fq_input_1" type="FakeQuantize" version="opset1">
+ <data auto_broadcast="numpy" levels="256"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1"/>
+ <port id="2"/>
+ <port id="3"/>
+ <port id="4"/>
+ </input>
+ <output>
+ <port id="5" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1614" name="bottleneck4_2/add" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_2/add" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1615" name="bottleneck4_2/fn/weights31076404931282" type="Const" version="opset1">
+ <data element_type="f32" offset="1576" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1616" name="bottleneck4_2/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_2/add" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1617" name="bottleneck4_3/add/fq_input_0" type="FakeQuantize" version="opset1">
+ <data auto_broadcast="numpy" levels="256"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1"/>
+ <port id="2"/>
+ <port id="3"/>
+ <port id="4"/>
+ </input>
+ <output>
+ <port id="5" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1618" name="4194419821612" type="Const" version="opset1">
+ <data element_type="f16" offset="280728" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="1619" name="4195419922626" type="Const" version="opset1">
+ <data element_type="f16" offset="280730" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="1620" name="4196420022434" type="Const" version="opset1">
+ <data element_type="f16" offset="280728" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="1621" name="4197420119560" type="Const" version="opset1">
+ <data element_type="f16" offset="280730" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="1622" name="3664366821177" type="Const" version="opset1">
+ <data element_type="f16" offset="280732" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="1623" name="3665366922953" type="Const" version="opset1">
+ <data element_type="f16" offset="280734" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="1624" name="3666367021291" type="Const" version="opset1">
+ <data element_type="f16" offset="280732" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="1625" name="3667367122464" type="Const" version="opset1">
+ <data element_type="f16" offset="280734" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="1626" name="4804480820217" type="Const" version="opset1">
+ <data element_type="f16" offset="280736" shape="1,64,1,1" size="128"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1627" name="4805480921045" type="Const" version="opset1">
+ <data element_type="f16" offset="280864" shape="1,64,1,1" size="128"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1628" name="4806481021951" type="Const" version="opset1">
+ <data element_type="f16" offset="280736" shape="1,64,1,1" size="128"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1629" name="4807481121279" type="Const" version="opset1">
+ <data element_type="f16" offset="280864" shape="1,64,1,1" size="128"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1630" name="bottleneck4_3/dim_red/bn/mean/Fused_Mul__copy128410331/quantized1328819668" type="Const" version="opset1">
+ <data element_type="i8" offset="280992" shape="64,256,1,1" size="16384"/>
+ <output>
+ <port id="0" precision="I8">
+ <dim>64</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1631" name="bottleneck4_3/dim_red/bn/mean/Fused_Mul__copy128410331/quantized/to_f16" type="Convert" version="opset1">
+ <data destination_type="f16"/>
+ <input>
+ <port id="0">
+ <dim>64</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="1" precision="FP16">
+ <dim>64</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1632" name="bottleneck4_3/dim_red/conv/fq_weights_1/zero_point1330121576" type="Const" version="opset1">
+ <data element_type="f16" offset="297376" shape="64,1,1,1" size="128"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1633" name="bottleneck4_3/dim_red/conv/fq_weights_1/minus_zp" type="Subtract" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>64</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="1">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>64</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1634" name="bottleneck4_3/dim_red/conv/fq_weights_1/scale1329622698" type="Const" version="opset1">
+ <data element_type="f16" offset="297504" shape="64,1,1,1" size="128"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1635" name="bottleneck4_3/dim_red/conv/fq_weights_1/mulpiply_by_scale" type="Multiply" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>64</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="1">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>64</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1636" name="bottleneck4_3/dim_red/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1,1" pads_begin="0,0" pads_end="0,0" strides="1,1"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1">
+ <dim>64</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1637" name="data_add_2432924334128620583" type="Const" version="opset1">
+ <data element_type="f16" offset="297632" shape="1,64,1,1" size="128"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1638" name="bottleneck4_3/dim_red/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_3/dim_red/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1639" name="bottleneck4_3/dim_red/fn/weights30932406851288" type="Const" version="opset1">
+ <data element_type="f32" offset="1576" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1640" name="bottleneck4_3/dim_red/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_3/dim_red/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1641" name="bottleneck4_3/inner/dw1/conv/fq_input_0" type="FakeQuantize" version="opset1">
+ <data auto_broadcast="numpy" levels="256"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="2">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="3">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="4">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="5" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1642" name="16903/value1690521288" type="Const" version="opset1">
+ <data element_type="i64" offset="189984" shape="5" size="40"/>
+ <output>
+ <port id="0" precision="I64">
+ <dim>5</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1643" name="bottleneck4_3/inner/dw1/bn/mean/Fused_Mul__copy129010334/quantized1201619842" type="Const" version="opset1">
+ <data element_type="i8" offset="297760" shape="64,1,3,3" size="576"/>
+ <output>
+ <port id="0" precision="I8">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1644" name="bottleneck4_3/inner/dw1/bn/mean/Fused_Mul__copy129010334/quantized/to_f16" type="Convert" version="opset1">
+ <data destination_type="f16"/>
+ <input>
+ <port id="0">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </input>
+ <output>
+ <port id="1" precision="FP16">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1645" name="bottleneck4_3/inner/dw1/conv/fq_weights_1/zero_point1202920493" type="Const" version="opset1">
+ <data element_type="f16" offset="298336" shape="64,1,1,1" size="128"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1646" name="bottleneck4_3/inner/dw1/conv/fq_weights_1/minus_zp" type="Subtract" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ <port id="1">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1647" name="bottleneck4_3/inner/dw1/conv/fq_weights_1/scale1202421861" type="Const" version="opset1">
+ <data element_type="f16" offset="298464" shape="64,1,1,1" size="128"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1648" name="bottleneck4_3/inner/dw1/conv/fq_weights_1/mulpiply_by_scale" type="Multiply" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ <port id="1">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1649" name="16903" type="Reshape" version="opset1">
+ <data special_zero="true"/>
+ <input>
+ <port id="0">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ <port id="1">
+ <dim>5</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1650" name="bottleneck4_3/inner/dw1/conv" type="GroupConvolution" version="opset1">
+ <data auto_pad="explicit" dilations="1,1" pads_begin="1,1" pads_end="1,1" strides="1,1"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1651" name="data_add_2433724342129221663" type="Const" version="opset1">
+ <data element_type="f16" offset="298592" shape="1,64,1,1" size="128"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1652" name="bottleneck4_3/inner/dw1/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_3/inner/dw1/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1653" name="bottleneck4_3/inner/dw1/fn/weights31084401301294" type="Const" version="opset1">
+ <data element_type="f32" offset="1576" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1654" name="bottleneck4_3/inner/dw1/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_3/inner/dw1/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1655" name="bottleneck4_3/dim_inc/conv/fq_input_0" type="FakeQuantize" version="opset1">
+ <data auto_broadcast="numpy" levels="256"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1"/>
+ <port id="2"/>
+ <port id="3"/>
+ <port id="4"/>
+ </input>
+ <output>
+ <port id="5" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1656" name="bottleneck4_3/dim_inc/bn/mean/Fused_Mul__copy129610337/quantized1345621945" type="Const" version="opset1">
+ <data element_type="i8" offset="298720" shape="256,64,1,1" size="16384"/>
+ <output>
+ <port id="0" precision="I8">
+ <dim>256</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1657" name="bottleneck4_3/dim_inc/bn/mean/Fused_Mul__copy129610337/quantized/to_f16" type="Convert" version="opset1">
+ <data destination_type="f16"/>
+ <input>
+ <port id="0">
+ <dim>256</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="1" precision="FP16">
+ <dim>256</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1658" name="bottleneck4_3/dim_inc/conv/fq_weights_1/zero_point1346922947" type="Const" version="opset1">
+ <data element_type="f16" offset="315104" shape="256,1,1,1" size="512"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1659" name="bottleneck4_3/dim_inc/conv/fq_weights_1/minus_zp" type="Subtract" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>256</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="1">
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>256</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1660" name="bottleneck4_3/dim_inc/conv/fq_weights_1/scale1346421705" type="Const" version="opset1">
+ <data element_type="f16" offset="315616" shape="256,1,1,1" size="512"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1661" name="bottleneck4_3/dim_inc/conv/fq_weights_1/mulpiply_by_scale" type="Multiply" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>256</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="1">
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>256</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1662" name="bottleneck4_3/dim_inc/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1,1" pads_begin="0,0" pads_end="0,0" strides="1,1"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1">
+ <dim>256</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1663" name="data_add_2434524350129820115" type="Const" version="opset1">
+ <data element_type="f16" offset="316128" shape="1,256,1,1" size="512"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1664" name="bottleneck4_3/dim_inc/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_3/dim_inc/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1665" name="bottleneck4_3/add/fq_input_1" type="FakeQuantize" version="opset1">
+ <data auto_broadcast="numpy" levels="256"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1"/>
+ <port id="2"/>
+ <port id="3"/>
+ <port id="4"/>
+ </input>
+ <output>
+ <port id="5" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1666" name="bottleneck4_3/add" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_3/add" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1667" name="bottleneck4_3/fn/weights30764404661301" type="Const" version="opset1">
+ <data element_type="f32" offset="1576" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1668" name="bottleneck4_3/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_3/add" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1669" name="bottleneck4_4/add/fq_input_0" type="FakeQuantize" version="opset1">
+ <data auto_broadcast="numpy" levels="256"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1"/>
+ <port id="2"/>
+ <port id="3"/>
+ <port id="4"/>
+ </input>
+ <output>
+ <port id="5" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1670" name="3554355822860" type="Const" version="opset1">
+ <data element_type="f16" offset="316640" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="1671" name="3555355922923" type="Const" version="opset1">
+ <data element_type="f16" offset="316642" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="1672" name="3556356021948" type="Const" version="opset1">
+ <data element_type="f16" offset="316640" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="1673" name="3557356121156" type="Const" version="opset1">
+ <data element_type="f16" offset="316642" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="1674" name="4364436821486" type="Const" version="opset1">
+ <data element_type="f16" offset="316644" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="1675" name="4365436920781" type="Const" version="opset1">
+ <data element_type="f16" offset="316646" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="1676" name="4366437019791" type="Const" version="opset1">
+ <data element_type="f16" offset="316644" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="1677" name="4367437119374" type="Const" version="opset1">
+ <data element_type="f16" offset="316646" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="1678" name="5504550820172" type="Const" version="opset1">
+ <data element_type="f16" offset="316648" shape="1,64,1,1" size="128"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1679" name="5505550920838" type="Const" version="opset1">
+ <data element_type="f16" offset="316776" shape="1,64,1,1" size="128"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1680" name="5506551019758" type="Const" version="opset1">
+ <data element_type="f16" offset="316648" shape="1,64,1,1" size="128"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1681" name="5507551121759" type="Const" version="opset1">
+ <data element_type="f16" offset="316776" shape="1,64,1,1" size="128"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1682" name="bottleneck4_4/dim_red/bn/mean/Fused_Mul__copy130310340/quantized1244820982" type="Const" version="opset1">
+ <data element_type="i8" offset="316904" shape="64,256,1,1" size="16384"/>
+ <output>
+ <port id="0" precision="I8">
+ <dim>64</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1683" name="bottleneck4_4/dim_red/bn/mean/Fused_Mul__copy130310340/quantized/to_f16" type="Convert" version="opset1">
+ <data destination_type="f16"/>
+ <input>
+ <port id="0">
+ <dim>64</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="1" precision="FP16">
+ <dim>64</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1684" name="bottleneck4_4/dim_red/conv/fq_weights_1/zero_point1246120223" type="Const" version="opset1">
+ <data element_type="f16" offset="333288" shape="64,1,1,1" size="128"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1685" name="bottleneck4_4/dim_red/conv/fq_weights_1/minus_zp" type="Subtract" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>64</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="1">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>64</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1686" name="bottleneck4_4/dim_red/conv/fq_weights_1/scale1245622836" type="Const" version="opset1">
+ <data element_type="f16" offset="333416" shape="64,1,1,1" size="128"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1687" name="bottleneck4_4/dim_red/conv/fq_weights_1/mulpiply_by_scale" type="Multiply" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>64</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="1">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>64</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1688" name="bottleneck4_4/dim_red/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1,1" pads_begin="0,0" pads_end="0,0" strides="1,1"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1">
+ <dim>64</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1689" name="data_add_2435324358130520409" type="Const" version="opset1">
+ <data element_type="f16" offset="333544" shape="1,64,1,1" size="128"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1690" name="bottleneck4_4/dim_red/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_4/dim_red/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1691" name="bottleneck4_4/dim_red/fn/weights30928400671307" type="Const" version="opset1">
+ <data element_type="f32" offset="1576" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1692" name="bottleneck4_4/dim_red/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_4/dim_red/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1693" name="bottleneck4_4/inner/dw1/conv/fq_input_0" type="FakeQuantize" version="opset1">
+ <data auto_broadcast="numpy" levels="256"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="2">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="3">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="4">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="5" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1694" name="16935/value1693722893" type="Const" version="opset1">
+ <data element_type="i64" offset="189984" shape="5" size="40"/>
+ <output>
+ <port id="0" precision="I64">
+ <dim>5</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1695" name="bottleneck4_4/inner/dw1/bn/mean/Fused_Mul__copy130910343/quantized1225621669" type="Const" version="opset1">
+ <data element_type="i8" offset="333672" shape="64,1,3,3" size="576"/>
+ <output>
+ <port id="0" precision="I8">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1696" name="bottleneck4_4/inner/dw1/bn/mean/Fused_Mul__copy130910343/quantized/to_f16" type="Convert" version="opset1">
+ <data destination_type="f16"/>
+ <input>
+ <port id="0">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </input>
+ <output>
+ <port id="1" precision="FP16">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1697" name="bottleneck4_4/inner/dw1/conv/fq_weights_1/zero_point1226919734" type="Const" version="opset1">
+ <data element_type="f16" offset="334248" shape="64,1,1,1" size="128"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1698" name="bottleneck4_4/inner/dw1/conv/fq_weights_1/minus_zp" type="Subtract" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ <port id="1">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1699" name="bottleneck4_4/inner/dw1/conv/fq_weights_1/scale1226419650" type="Const" version="opset1">
+ <data element_type="f16" offset="334376" shape="64,1,1,1" size="128"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1700" name="bottleneck4_4/inner/dw1/conv/fq_weights_1/mulpiply_by_scale" type="Multiply" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ <port id="1">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1701" name="16935" type="Reshape" version="opset1">
+ <data special_zero="true"/>
+ <input>
+ <port id="0">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ <port id="1">
+ <dim>5</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1702" name="bottleneck4_4/inner/dw1/conv" type="GroupConvolution" version="opset1">
+ <data auto_pad="explicit" dilations="1,1" pads_begin="1,1" pads_end="1,1" strides="1,1"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1703" name="data_add_2436124366131121252" type="Const" version="opset1">
+ <data element_type="f16" offset="334504" shape="1,64,1,1" size="128"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1704" name="bottleneck4_4/inner/dw1/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_4/inner/dw1/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1705" name="bottleneck4_4/inner/dw1/fn/weights30992404541313" type="Const" version="opset1">
+ <data element_type="f32" offset="1576" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1706" name="bottleneck4_4/inner/dw1/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_4/inner/dw1/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1707" name="bottleneck4_4/dim_inc/conv/fq_input_0" type="FakeQuantize" version="opset1">
+ <data auto_broadcast="numpy" levels="256"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1"/>
+ <port id="2"/>
+ <port id="3"/>
+ <port id="4"/>
+ </input>
+ <output>
+ <port id="5" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1708" name="bottleneck4_4/dim_inc/bn/mean/Fused_Mul__copy131510346/quantized1415221723" type="Const" version="opset1">
+ <data element_type="i8" offset="334632" shape="256,64,1,1" size="16384"/>
+ <output>
+ <port id="0" precision="I8">
+ <dim>256</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1709" name="bottleneck4_4/dim_inc/bn/mean/Fused_Mul__copy131510346/quantized/to_f16" type="Convert" version="opset1">
+ <data destination_type="f16"/>
+ <input>
+ <port id="0">
+ <dim>256</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="1" precision="FP16">
+ <dim>256</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1710" name="bottleneck4_4/dim_inc/conv/fq_weights_1/zero_point1416521732" type="Const" version="opset1">
+ <data element_type="f16" offset="351016" shape="256,1,1,1" size="512"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1711" name="bottleneck4_4/dim_inc/conv/fq_weights_1/minus_zp" type="Subtract" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>256</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="1">
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>256</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1712" name="bottleneck4_4/dim_inc/conv/fq_weights_1/scale1416022620" type="Const" version="opset1">
+ <data element_type="f16" offset="351528" shape="256,1,1,1" size="512"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1713" name="bottleneck4_4/dim_inc/conv/fq_weights_1/mulpiply_by_scale" type="Multiply" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>256</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="1">
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>256</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1714" name="bottleneck4_4/dim_inc/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1,1" pads_begin="0,0" pads_end="0,0" strides="1,1"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1">
+ <dim>256</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1715" name="data_add_2436924374131720943" type="Const" version="opset1">
+ <data element_type="f16" offset="352040" shape="1,256,1,1" size="512"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1716" name="bottleneck4_4/dim_inc/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_4/dim_inc/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1717" name="bottleneck4_4/add/fq_input_1" type="FakeQuantize" version="opset1">
+ <data auto_broadcast="numpy" levels="256"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1"/>
+ <port id="2"/>
+ <port id="3"/>
+ <port id="4"/>
+ </input>
+ <output>
+ <port id="5" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1718" name="bottleneck4_4/add" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_4/add" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1719" name="bottleneck4_4/fn/weights30996399261320" type="Const" version="opset1">
+ <data element_type="f32" offset="1576" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1720" name="bottleneck4_4/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_4/add" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1721" name="bottleneck4_5/add/fq_input_0" type="FakeQuantize" version="opset1">
+ <data auto_broadcast="numpy" levels="256"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1"/>
+ <port id="2"/>
+ <port id="3"/>
+ <port id="4"/>
+ </input>
+ <output>
+ <port id="5" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1722" name="4274427822878" type="Const" version="opset1">
+ <data element_type="f16" offset="352552" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="1723" name="4275427919563" type="Const" version="opset1">
+ <data element_type="f16" offset="352554" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="1724" name="4276428021600" type="Const" version="opset1">
+ <data element_type="f16" offset="352552" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="1725" name="4277428121873" type="Const" version="opset1">
+ <data element_type="f16" offset="352554" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="1726" name="3884388820109" type="Const" version="opset1">
+ <data element_type="f16" offset="352556" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="1727" name="3885388921066" type="Const" version="opset1">
+ <data element_type="f16" offset="352558" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="1728" name="3886389022656" type="Const" version="opset1">
+ <data element_type="f16" offset="352556" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="1729" name="3887389120121" type="Const" version="opset1">
+ <data element_type="f16" offset="352558" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="1730" name="4424442821330" type="Const" version="opset1">
+ <data element_type="f16" offset="352560" shape="1,64,1,1" size="128"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1731" name="4425442919746" type="Const" version="opset1">
+ <data element_type="f16" offset="352688" shape="1,64,1,1" size="128"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1732" name="4426443020889" type="Const" version="opset1">
+ <data element_type="f16" offset="352560" shape="1,64,1,1" size="128"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1733" name="4427443120718" type="Const" version="opset1">
+ <data element_type="f16" offset="352688" shape="1,64,1,1" size="128"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1734" name="bottleneck4_5/dim_red/bn/mean/Fused_Mul__copy132210349/quantized1412822845" type="Const" version="opset1">
+ <data element_type="i8" offset="352816" shape="64,256,1,1" size="16384"/>
+ <output>
+ <port id="0" precision="I8">
+ <dim>64</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1735" name="bottleneck4_5/dim_red/bn/mean/Fused_Mul__copy132210349/quantized/to_f16" type="Convert" version="opset1">
+ <data destination_type="f16"/>
+ <input>
+ <port id="0">
+ <dim>64</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="1" precision="FP16">
+ <dim>64</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1736" name="bottleneck4_5/dim_red/conv/fq_weights_1/zero_point1414121765" type="Const" version="opset1">
+ <data element_type="f16" offset="369200" shape="64,1,1,1" size="128"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1737" name="bottleneck4_5/dim_red/conv/fq_weights_1/minus_zp" type="Subtract" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>64</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="1">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>64</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1738" name="bottleneck4_5/dim_red/conv/fq_weights_1/scale1413622977" type="Const" version="opset1">
+ <data element_type="f16" offset="369328" shape="64,1,1,1" size="128"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1739" name="bottleneck4_5/dim_red/conv/fq_weights_1/mulpiply_by_scale" type="Multiply" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>64</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="1">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>64</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1740" name="bottleneck4_5/dim_red/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1,1" pads_begin="0,0" pads_end="0,0" strides="1,1"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1">
+ <dim>64</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1741" name="data_add_2437724382132419416" type="Const" version="opset1">
+ <data element_type="f16" offset="369456" shape="1,64,1,1" size="128"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1742" name="bottleneck4_5/dim_red/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_5/dim_red/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1743" name="bottleneck4_5/dim_red/fn/weights30804406821326" type="Const" version="opset1">
+ <data element_type="f32" offset="1576" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1744" name="bottleneck4_5/dim_red/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_5/dim_red/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1745" name="bottleneck4_5/inner/dw1/conv/fq_input_0" type="FakeQuantize" version="opset1">
+ <data auto_broadcast="numpy" levels="256"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="2">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="3">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="4">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="5" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1746" name="16891/value1689321123" type="Const" version="opset1">
+ <data element_type="i64" offset="189984" shape="5" size="40"/>
+ <output>
+ <port id="0" precision="I64">
+ <dim>5</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1747" name="bottleneck4_5/inner/dw1/bn/mean/Fused_Mul__copy132810352/quantized1372020865" type="Const" version="opset1">
+ <data element_type="i8" offset="369584" shape="64,1,3,3" size="576"/>
+ <output>
+ <port id="0" precision="I8">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1748" name="bottleneck4_5/inner/dw1/bn/mean/Fused_Mul__copy132810352/quantized/to_f16" type="Convert" version="opset1">
+ <data destination_type="f16"/>
+ <input>
+ <port id="0">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </input>
+ <output>
+ <port id="1" precision="FP16">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1749" name="bottleneck4_5/inner/dw1/conv/fq_weights_1/zero_point1373320262" type="Const" version="opset1">
+ <data element_type="f16" offset="370160" shape="64,1,1,1" size="128"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1750" name="bottleneck4_5/inner/dw1/conv/fq_weights_1/minus_zp" type="Subtract" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ <port id="1">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1751" name="bottleneck4_5/inner/dw1/conv/fq_weights_1/scale1372821525" type="Const" version="opset1">
+ <data element_type="f16" offset="370288" shape="64,1,1,1" size="128"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1752" name="bottleneck4_5/inner/dw1/conv/fq_weights_1/mulpiply_by_scale" type="Multiply" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ <port id="1">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1753" name="16891" type="Reshape" version="opset1">
+ <data special_zero="true"/>
+ <input>
+ <port id="0">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ <port id="1">
+ <dim>5</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1754" name="bottleneck4_5/inner/dw1/conv" type="GroupConvolution" version="opset1">
+ <data auto_pad="explicit" dilations="1,1" pads_begin="1,1" pads_end="1,1" strides="1,1"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1755" name="data_add_2438524390133022614" type="Const" version="opset1">
+ <data element_type="f16" offset="370416" shape="1,64,1,1" size="128"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1756" name="bottleneck4_5/inner/dw1/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_5/inner/dw1/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1757" name="bottleneck4_5/inner/dw1/fn/weights30900396801332" type="Const" version="opset1">
+ <data element_type="f32" offset="1576" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1758" name="bottleneck4_5/inner/dw1/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_5/inner/dw1/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1759" name="bottleneck4_5/dim_inc/conv/fq_input_0" type="FakeQuantize" version="opset1">
+ <data auto_broadcast="numpy" levels="256"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1"/>
+ <port id="2"/>
+ <port id="3"/>
+ <port id="4"/>
+ </input>
+ <output>
+ <port id="5" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1760" name="bottleneck4_5/dim_inc/bn/mean/Fused_Mul__copy133410355/quantized1340820727" type="Const" version="opset1">
+ <data element_type="i8" offset="370544" shape="256,64,1,1" size="16384"/>
+ <output>
+ <port id="0" precision="I8">
+ <dim>256</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1761" name="bottleneck4_5/dim_inc/bn/mean/Fused_Mul__copy133410355/quantized/to_f16" type="Convert" version="opset1">
+ <data destination_type="f16"/>
+ <input>
+ <port id="0">
+ <dim>256</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="1" precision="FP16">
+ <dim>256</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1762" name="bottleneck4_5/dim_inc/conv/fq_weights_1/zero_point1342120445" type="Const" version="opset1">
+ <data element_type="f16" offset="386928" shape="256,1,1,1" size="512"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1763" name="bottleneck4_5/dim_inc/conv/fq_weights_1/minus_zp" type="Subtract" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>256</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="1">
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>256</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1764" name="bottleneck4_5/dim_inc/conv/fq_weights_1/scale1341620247" type="Const" version="opset1">
+ <data element_type="f16" offset="387440" shape="256,1,1,1" size="512"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1765" name="bottleneck4_5/dim_inc/conv/fq_weights_1/mulpiply_by_scale" type="Multiply" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>256</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="1">
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>256</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1766" name="bottleneck4_5/dim_inc/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1,1" pads_begin="0,0" pads_end="0,0" strides="1,1"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1">
+ <dim>256</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1767" name="data_add_2439324398133619437" type="Const" version="opset1">
+ <data element_type="f16" offset="387952" shape="1,256,1,1" size="512"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1768" name="bottleneck4_5/dim_inc/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_5/dim_inc/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1769" name="bottleneck4_5/add/fq_input_1" type="FakeQuantize" version="opset1">
+ <data auto_broadcast="numpy" levels="256"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1"/>
+ <port id="2"/>
+ <port id="3"/>
+ <port id="4"/>
+ </input>
+ <output>
+ <port id="5" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1770" name="bottleneck4_5/add" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_5/add" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1771" name="bottleneck4_5/fn/weights30876398571339" type="Const" version="opset1">
+ <data element_type="f32" offset="1576" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1772" name="bottleneck4_5/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_5/add" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1773" name="bottleneck4_6/add/fq_input_0" type="FakeQuantize" version="opset1">
+ <data auto_broadcast="numpy" levels="256"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1"/>
+ <port id="2"/>
+ <port id="3"/>
+ <port id="4"/>
+ </input>
+ <output>
+ <port id="5" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1774" name="3654365819443" type="Const" version="opset1">
+ <data element_type="f16" offset="388464" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="1775" name="3655365922563" type="Const" version="opset1">
+ <data element_type="f16" offset="388466" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="1776" name="3656366021522" type="Const" version="opset1">
+ <data element_type="f16" offset="388464" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="1777" name="3657366121603" type="Const" version="opset1">
+ <data element_type="f16" offset="388466" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="1778" name="5224522821594" type="Const" version="opset1">
+ <data element_type="f16" offset="388468" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="1779" name="5225522920526" type="Const" version="opset1">
+ <data element_type="f16" offset="388470" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="1780" name="5226523022527" type="Const" version="opset1">
+ <data element_type="f16" offset="388468" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="1781" name="5227523120157" type="Const" version="opset1">
+ <data element_type="f16" offset="388470" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="1782" name="4304430822476" type="Const" version="opset1">
+ <data element_type="f16" offset="388472" shape="1,64,1,1" size="128"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1783" name="4305430921246" type="Const" version="opset1">
+ <data element_type="f16" offset="388600" shape="1,64,1,1" size="128"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1784" name="4306431021162" type="Const" version="opset1">
+ <data element_type="f16" offset="388472" shape="1,64,1,1" size="128"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1785" name="4307431122680" type="Const" version="opset1">
+ <data element_type="f16" offset="388600" shape="1,64,1,1" size="128"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1786" name="bottleneck4_6/dim_red/bn/mean/Fused_Mul__copy134110358/quantized1417620322" type="Const" version="opset1">
+ <data element_type="i8" offset="388728" shape="64,256,1,1" size="16384"/>
+ <output>
+ <port id="0" precision="I8">
+ <dim>64</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1787" name="bottleneck4_6/dim_red/bn/mean/Fused_Mul__copy134110358/quantized/to_f16" type="Convert" version="opset1">
+ <data destination_type="f16"/>
+ <input>
+ <port id="0">
+ <dim>64</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="1" precision="FP16">
+ <dim>64</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1788" name="bottleneck4_6/dim_red/conv/fq_weights_1/zero_point1418922356" type="Const" version="opset1">
+ <data element_type="f16" offset="405112" shape="64,1,1,1" size="128"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1789" name="bottleneck4_6/dim_red/conv/fq_weights_1/minus_zp" type="Subtract" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>64</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="1">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>64</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1790" name="bottleneck4_6/dim_red/conv/fq_weights_1/scale1418421513" type="Const" version="opset1">
+ <data element_type="f16" offset="405240" shape="64,1,1,1" size="128"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1791" name="bottleneck4_6/dim_red/conv/fq_weights_1/mulpiply_by_scale" type="Multiply" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>64</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="1">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>64</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1792" name="bottleneck4_6/dim_red/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1,1" pads_begin="0,0" pads_end="0,0" strides="1,1"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1">
+ <dim>64</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1793" name="data_add_2440124406134320532" type="Const" version="opset1">
+ <data element_type="f16" offset="405368" shape="1,64,1,1" size="128"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1794" name="bottleneck4_6/dim_red/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_6/dim_red/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1795" name="bottleneck4_6/dim_red/fn/weights30864401721345" type="Const" version="opset1">
+ <data element_type="f32" offset="1576" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1796" name="bottleneck4_6/dim_red/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_6/dim_red/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1797" name="bottleneck4_6/inner/dw1/conv/fq_input_0" type="FakeQuantize" version="opset1">
+ <data auto_broadcast="numpy" levels="256"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="2">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="3">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="4">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="5" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1798" name="16883/value1688519941" type="Const" version="opset1">
+ <data element_type="i64" offset="189984" shape="5" size="40"/>
+ <output>
+ <port id="0" precision="I64">
+ <dim>5</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1799" name="bottleneck4_6/inner/dw1/bn/mean/Fused_Mul__copy134710361/quantized1331220829" type="Const" version="opset1">
+ <data element_type="i8" offset="405496" shape="64,1,3,3" size="576"/>
+ <output>
+ <port id="0" precision="I8">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1800" name="bottleneck4_6/inner/dw1/bn/mean/Fused_Mul__copy134710361/quantized/to_f16" type="Convert" version="opset1">
+ <data destination_type="f16"/>
+ <input>
+ <port id="0">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </input>
+ <output>
+ <port id="1" precision="FP16">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1801" name="bottleneck4_6/inner/dw1/conv/fq_weights_1/zero_point1332522011" type="Const" version="opset1">
+ <data element_type="f16" offset="406072" shape="64,1,1,1" size="128"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1802" name="bottleneck4_6/inner/dw1/conv/fq_weights_1/minus_zp" type="Subtract" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ <port id="1">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1803" name="bottleneck4_6/inner/dw1/conv/fq_weights_1/scale1332021300" type="Const" version="opset1">
+ <data element_type="f16" offset="406200" shape="64,1,1,1" size="128"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1804" name="bottleneck4_6/inner/dw1/conv/fq_weights_1/mulpiply_by_scale" type="Multiply" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ <port id="1">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1805" name="16883" type="Reshape" version="opset1">
+ <data special_zero="true"/>
+ <input>
+ <port id="0">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ <port id="1">
+ <dim>5</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1806" name="bottleneck4_6/inner/dw1/conv" type="GroupConvolution" version="opset1">
+ <data auto_pad="explicit" dilations="1,1" pads_begin="1,1" pads_end="1,1" strides="1,1"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1807" name="data_add_2440924414134921048" type="Const" version="opset1">
+ <data element_type="f16" offset="406328" shape="1,64,1,1" size="128"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1808" name="bottleneck4_6/inner/dw1/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_6/inner/dw1/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1809" name="bottleneck4_6/inner/dw1/fn/weights31116402381351" type="Const" version="opset1">
+ <data element_type="f32" offset="1576" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1810" name="bottleneck4_6/inner/dw1/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_6/inner/dw1/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1811" name="bottleneck4_6/dim_inc/conv/fq_input_0" type="FakeQuantize" version="opset1">
+ <data auto_broadcast="numpy" levels="256"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1"/>
+ <port id="2"/>
+ <port id="3"/>
+ <port id="4"/>
+ </input>
+ <output>
+ <port id="5" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1812" name="bottleneck4_6/dim_inc/bn/mean/Fused_Mul__copy135310364/quantized1381619932" type="Const" version="opset1">
+ <data element_type="i8" offset="406456" shape="256,64,1,1" size="16384"/>
+ <output>
+ <port id="0" precision="I8">
+ <dim>256</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1813" name="bottleneck4_6/dim_inc/bn/mean/Fused_Mul__copy135310364/quantized/to_f16" type="Convert" version="opset1">
+ <data destination_type="f16"/>
+ <input>
+ <port id="0">
+ <dim>256</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="1" precision="FP16">
+ <dim>256</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1814" name="bottleneck4_6/dim_inc/conv/fq_weights_1/zero_point1382919689" type="Const" version="opset1">
+ <data element_type="f16" offset="422840" shape="256,1,1,1" size="512"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1815" name="bottleneck4_6/dim_inc/conv/fq_weights_1/minus_zp" type="Subtract" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>256</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="1">
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>256</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1816" name="bottleneck4_6/dim_inc/conv/fq_weights_1/scale1382422968" type="Const" version="opset1">
+ <data element_type="f16" offset="423352" shape="256,1,1,1" size="512"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1817" name="bottleneck4_6/dim_inc/conv/fq_weights_1/mulpiply_by_scale" type="Multiply" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>256</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="1">
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>256</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1818" name="bottleneck4_6/dim_inc/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1,1" pads_begin="0,0" pads_end="0,0" strides="1,1"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1">
+ <dim>256</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1819" name="data_add_2441724422135519569" type="Const" version="opset1">
+ <data element_type="f16" offset="423864" shape="1,256,1,1" size="512"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1820" name="bottleneck4_6/dim_inc/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_6/dim_inc/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1821" name="bottleneck4_6/add/fq_input_1" type="FakeQuantize" version="opset1">
+ <data auto_broadcast="numpy" levels="256"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1"/>
+ <port id="2"/>
+ <port id="3"/>
+ <port id="4"/>
+ </input>
+ <output>
+ <port id="5" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1822" name="bottleneck4_6/add" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_6/add" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1823" name="bottleneck4_6/fn/weights30820397731358" type="Const" version="opset1">
+ <data element_type="f32" offset="1576" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1824" name="bottleneck4_6/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_6/add" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1825" name="bottleneck4_7/add/fq_input_0" type="FakeQuantize" version="opset1">
+ <data auto_broadcast="numpy" levels="256"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1"/>
+ <port id="2"/>
+ <port id="3"/>
+ <port id="4"/>
+ </input>
+ <output>
+ <port id="5" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1826" name="2714271821972" type="Const" version="opset1">
+ <data element_type="f16" offset="424376" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="1827" name="2715271922662" type="Const" version="opset1">
+ <data element_type="f16" offset="424378" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="1828" name="2716272020352" type="Const" version="opset1">
+ <data element_type="f16" offset="424376" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="1829" name="2717272122233" type="Const" version="opset1">
+ <data element_type="f16" offset="424378" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="1830" name="4924492820382" type="Const" version="opset1">
+ <data element_type="f16" offset="424380" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="1831" name="4925492922206" type="Const" version="opset1">
+ <data element_type="f16" offset="424382" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="1832" name="4926493020565" type="Const" version="opset1">
+ <data element_type="f16" offset="424380" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="1833" name="4927493119953" type="Const" version="opset1">
+ <data element_type="f16" offset="424382" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="1834" name="3264326822242" type="Const" version="opset1">
+ <data element_type="f16" offset="424384" shape="1,64,1,1" size="128"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1835" name="3265326921876" type="Const" version="opset1">
+ <data element_type="f16" offset="424512" shape="1,64,1,1" size="128"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1836" name="3266327022854" type="Const" version="opset1">
+ <data element_type="f16" offset="424384" shape="1,64,1,1" size="128"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1837" name="3267327119479" type="Const" version="opset1">
+ <data element_type="f16" offset="424512" shape="1,64,1,1" size="128"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1838" name="bottleneck4_7/dim_red/bn/mean/Fused_Mul__copy136010367/quantized1396019887" type="Const" version="opset1">
+ <data element_type="i8" offset="424640" shape="64,256,1,1" size="16384"/>
+ <output>
+ <port id="0" precision="I8">
+ <dim>64</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1839" name="bottleneck4_7/dim_red/bn/mean/Fused_Mul__copy136010367/quantized/to_f16" type="Convert" version="opset1">
+ <data destination_type="f16"/>
+ <input>
+ <port id="0">
+ <dim>64</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="1" precision="FP16">
+ <dim>64</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1840" name="bottleneck4_7/dim_red/conv/fq_weights_1/zero_point1397320340" type="Const" version="opset1">
+ <data element_type="f16" offset="441024" shape="64,1,1,1" size="128"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1841" name="bottleneck4_7/dim_red/conv/fq_weights_1/minus_zp" type="Subtract" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>64</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="1">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>64</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1842" name="bottleneck4_7/dim_red/conv/fq_weights_1/scale1396821975" type="Const" version="opset1">
+ <data element_type="f16" offset="441152" shape="64,1,1,1" size="128"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1843" name="bottleneck4_7/dim_red/conv/fq_weights_1/mulpiply_by_scale" type="Multiply" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>64</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="1">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>64</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1844" name="bottleneck4_7/dim_red/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1,1" pads_begin="0,0" pads_end="0,0" strides="1,1"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1">
+ <dim>64</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1845" name="data_add_2442524430136219935" type="Const" version="opset1">
+ <data element_type="f16" offset="441280" shape="1,64,1,1" size="128"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1846" name="bottleneck4_7/dim_red/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_7/dim_red/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1847" name="bottleneck4_7/dim_red/fn/weights30816401931364" type="Const" version="opset1">
+ <data element_type="f32" offset="1576" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1848" name="bottleneck4_7/dim_red/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_7/dim_red/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1849" name="bottleneck4_7/inner/dw1/conv/fq_input_0" type="FakeQuantize" version="opset1">
+ <data auto_broadcast="numpy" levels="256"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="2">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="3">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="4">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="5" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1850" name="16823/value1682521726" type="Const" version="opset1">
+ <data element_type="i64" offset="189984" shape="5" size="40"/>
+ <output>
+ <port id="0" precision="I64">
+ <dim>5</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1851" name="bottleneck4_7/inner/dw1/bn/mean/Fused_Mul__copy136610370/quantized1302419512" type="Const" version="opset1">
+ <data element_type="i8" offset="441408" shape="64,1,3,3" size="576"/>
+ <output>
+ <port id="0" precision="I8">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1852" name="bottleneck4_7/inner/dw1/bn/mean/Fused_Mul__copy136610370/quantized/to_f16" type="Convert" version="opset1">
+ <data destination_type="f16"/>
+ <input>
+ <port id="0">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </input>
+ <output>
+ <port id="1" precision="FP16">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1853" name="bottleneck4_7/inner/dw1/conv/fq_weights_1/zero_point1303720316" type="Const" version="opset1">
+ <data element_type="f16" offset="441984" shape="64,1,1,1" size="128"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1854" name="bottleneck4_7/inner/dw1/conv/fq_weights_1/minus_zp" type="Subtract" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ <port id="1">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1855" name="bottleneck4_7/inner/dw1/conv/fq_weights_1/scale1303222362" type="Const" version="opset1">
+ <data element_type="f16" offset="442112" shape="64,1,1,1" size="128"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1856" name="bottleneck4_7/inner/dw1/conv/fq_weights_1/mulpiply_by_scale" type="Multiply" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ <port id="1">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1857" name="16823" type="Reshape" version="opset1">
+ <data special_zero="true"/>
+ <input>
+ <port id="0">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ <port id="1">
+ <dim>5</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1858" name="bottleneck4_7/inner/dw1/conv" type="GroupConvolution" version="opset1">
+ <data auto_pad="explicit" dilations="1,1" pads_begin="1,1" pads_end="1,1" strides="1,1"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1859" name="data_add_2443324438136819881" type="Const" version="opset1">
+ <data element_type="f16" offset="442240" shape="1,64,1,1" size="128"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1860" name="bottleneck4_7/inner/dw1/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_7/inner/dw1/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1861" name="bottleneck4_7/inner/dw1/fn/weights31112402261370" type="Const" version="opset1">
+ <data element_type="f32" offset="1576" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1862" name="bottleneck4_7/inner/dw1/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_7/inner/dw1/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1863" name="bottleneck4_7/dim_inc/conv/fq_input_0" type="FakeQuantize" version="opset1">
+ <data auto_broadcast="numpy" levels="256"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1"/>
+ <port id="2"/>
+ <port id="3"/>
+ <port id="4"/>
+ </input>
+ <output>
+ <port id="5" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1864" name="bottleneck4_7/dim_inc/bn/mean/Fused_Mul__copy137210373/quantized1271220460" type="Const" version="opset1">
+ <data element_type="i8" offset="442368" shape="256,64,1,1" size="16384"/>
+ <output>
+ <port id="0" precision="I8">
+ <dim>256</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1865" name="bottleneck4_7/dim_inc/bn/mean/Fused_Mul__copy137210373/quantized/to_f16" type="Convert" version="opset1">
+ <data destination_type="f16"/>
+ <input>
+ <port id="0">
+ <dim>256</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="1" precision="FP16">
+ <dim>256</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1866" name="bottleneck4_7/dim_inc/conv/fq_weights_1/zero_point1272520853" type="Const" version="opset1">
+ <data element_type="f16" offset="458752" shape="256,1,1,1" size="512"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1867" name="bottleneck4_7/dim_inc/conv/fq_weights_1/minus_zp" type="Subtract" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>256</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="1">
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>256</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1868" name="bottleneck4_7/dim_inc/conv/fq_weights_1/scale1272022896" type="Const" version="opset1">
+ <data element_type="f16" offset="459264" shape="256,1,1,1" size="512"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1869" name="bottleneck4_7/dim_inc/conv/fq_weights_1/mulpiply_by_scale" type="Multiply" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>256</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="1">
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>256</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1870" name="bottleneck4_7/dim_inc/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1,1" pads_begin="0,0" pads_end="0,0" strides="1,1"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1">
+ <dim>256</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1871" name="data_add_2444124446137421471" type="Const" version="opset1">
+ <data element_type="f16" offset="459776" shape="1,256,1,1" size="512"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1872" name="bottleneck4_7/dim_inc/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_7/dim_inc/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1873" name="bottleneck4_7/add/fq_input_1" type="FakeQuantize" version="opset1">
+ <data auto_broadcast="numpy" levels="256"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1"/>
+ <port id="2"/>
+ <port id="3"/>
+ <port id="4"/>
+ </input>
+ <output>
+ <port id="5" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1874" name="bottleneck4_7/add" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_7/add" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1875" name="bottleneck4_7/fn/weights31088405411377" type="Const" version="opset1">
+ <data element_type="f32" offset="1576" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1876" name="bottleneck4_7/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_7/add" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1877" name="bottleneck4_8/add/fq_input_0" type="FakeQuantize" version="opset1">
+ <data auto_broadcast="numpy" levels="256"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1"/>
+ <port id="2"/>
+ <port id="3"/>
+ <port id="4"/>
+ </input>
+ <output>
+ <port id="5" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1878" name="5094509822215" type="Const" version="opset1">
+ <data element_type="f16" offset="460288" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="1879" name="5095509922593" type="Const" version="opset1">
+ <data element_type="f16" offset="460290" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="1880" name="5096510021438" type="Const" version="opset1">
+ <data element_type="f16" offset="460288" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="1881" name="5097510119770" type="Const" version="opset1">
+ <data element_type="f16" offset="460290" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="1882" name="3624362821117" type="Const" version="opset1">
+ <data element_type="f16" offset="460292" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="1883" name="3625362921543" type="Const" version="opset1">
+ <data element_type="f16" offset="460294" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="1884" name="3626363022023" type="Const" version="opset1">
+ <data element_type="f16" offset="460292" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="1885" name="3627363122647" type="Const" version="opset1">
+ <data element_type="f16" offset="460294" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="1886" name="3444344820952" type="Const" version="opset1">
+ <data element_type="f16" offset="460296" shape="1,64,1,1" size="128"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1887" name="3445344921621" type="Const" version="opset1">
+ <data element_type="f16" offset="460424" shape="1,64,1,1" size="128"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1888" name="3446345022500" type="Const" version="opset1">
+ <data element_type="f16" offset="460296" shape="1,64,1,1" size="128"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1889" name="3447345122788" type="Const" version="opset1">
+ <data element_type="f16" offset="460424" shape="1,64,1,1" size="128"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1890" name="bottleneck4_8/dim_red/bn/mean/Fused_Mul__copy137910376/quantized1192020373" type="Const" version="opset1">
+ <data element_type="i8" offset="460552" shape="64,256,1,1" size="16384"/>
+ <output>
+ <port id="0" precision="I8">
+ <dim>64</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1891" name="bottleneck4_8/dim_red/bn/mean/Fused_Mul__copy137910376/quantized/to_f16" type="Convert" version="opset1">
+ <data destination_type="f16"/>
+ <input>
+ <port id="0">
+ <dim>64</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="1" precision="FP16">
+ <dim>64</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1892" name="bottleneck4_8/dim_red/conv/fq_weights_1/zero_point1193321264" type="Const" version="opset1">
+ <data element_type="f16" offset="476936" shape="64,1,1,1" size="128"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1893" name="bottleneck4_8/dim_red/conv/fq_weights_1/minus_zp" type="Subtract" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>64</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="1">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>64</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1894" name="bottleneck4_8/dim_red/conv/fq_weights_1/scale1192820028" type="Const" version="opset1">
+ <data element_type="f16" offset="477064" shape="64,1,1,1" size="128"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1895" name="bottleneck4_8/dim_red/conv/fq_weights_1/mulpiply_by_scale" type="Multiply" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>64</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="1">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>64</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1896" name="bottleneck4_8/dim_red/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1,1" pads_begin="0,0" pads_end="0,0" strides="1,1"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1">
+ <dim>64</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1897" name="data_add_2444924454138121021" type="Const" version="opset1">
+ <data element_type="f16" offset="477192" shape="1,64,1,1" size="128"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1898" name="bottleneck4_8/dim_red/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_8/dim_red/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1899" name="bottleneck4_8/dim_red/fn/weights31128404361383" type="Const" version="opset1">
+ <data element_type="f32" offset="1576" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1900" name="bottleneck4_8/dim_red/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_8/dim_red/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1901" name="bottleneck4_8/inner/dw1/conv/fq_input_0" type="FakeQuantize" version="opset1">
+ <data auto_broadcast="numpy" levels="256"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="2">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="3">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="4">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="5" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1902" name="16835/value1683721348" type="Const" version="opset1">
+ <data element_type="i64" offset="189984" shape="5" size="40"/>
+ <output>
+ <port id="0" precision="I64">
+ <dim>5</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1903" name="bottleneck4_8/inner/dw1/bn/mean/Fused_Mul__copy138510379/quantized1199221804" type="Const" version="opset1">
+ <data element_type="i8" offset="477320" shape="64,1,3,3" size="576"/>
+ <output>
+ <port id="0" precision="I8">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1904" name="bottleneck4_8/inner/dw1/bn/mean/Fused_Mul__copy138510379/quantized/to_f16" type="Convert" version="opset1">
+ <data destination_type="f16"/>
+ <input>
+ <port id="0">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </input>
+ <output>
+ <port id="1" precision="FP16">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1905" name="bottleneck4_8/inner/dw1/conv/fq_weights_1/zero_point1200522617" type="Const" version="opset1">
+ <data element_type="f16" offset="477896" shape="64,1,1,1" size="128"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1906" name="bottleneck4_8/inner/dw1/conv/fq_weights_1/minus_zp" type="Subtract" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ <port id="1">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1907" name="bottleneck4_8/inner/dw1/conv/fq_weights_1/scale1200022920" type="Const" version="opset1">
+ <data element_type="f16" offset="478024" shape="64,1,1,1" size="128"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1908" name="bottleneck4_8/inner/dw1/conv/fq_weights_1/mulpiply_by_scale" type="Multiply" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ <port id="1">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1909" name="16835" type="Reshape" version="opset1">
+ <data special_zero="true"/>
+ <input>
+ <port id="0">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ <port id="1">
+ <dim>5</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1910" name="bottleneck4_8/inner/dw1/conv" type="GroupConvolution" version="opset1">
+ <data auto_pad="explicit" dilations="1,1" pads_begin="1,1" pads_end="1,1" strides="1,1"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1911" name="data_add_2445724462138720841" type="Const" version="opset1">
+ <data element_type="f16" offset="478152" shape="1,64,1,1" size="128"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1912" name="bottleneck4_8/inner/dw1/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_8/inner/dw1/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1913" name="bottleneck4_8/inner/dw1/fn/weights30888398541389" type="Const" version="opset1">
+ <data element_type="f32" offset="1576" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1914" name="bottleneck4_8/inner/dw1/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_8/inner/dw1/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1915" name="bottleneck4_8/dim_inc/conv/fq_input_0" type="FakeQuantize" version="opset1">
+ <data auto_broadcast="numpy" levels="256"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1"/>
+ <port id="2"/>
+ <port id="3"/>
+ <port id="4"/>
+ </input>
+ <output>
+ <port id="5" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1916" name="bottleneck4_8/dim_inc/bn/mean/Fused_Mul__copy139110382/quantized1223221762" type="Const" version="opset1">
+ <data element_type="i8" offset="478280" shape="256,64,1,1" size="16384"/>
+ <output>
+ <port id="0" precision="I8">
+ <dim>256</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1917" name="bottleneck4_8/dim_inc/bn/mean/Fused_Mul__copy139110382/quantized/to_f16" type="Convert" version="opset1">
+ <data destination_type="f16"/>
+ <input>
+ <port id="0">
+ <dim>256</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="1" precision="FP16">
+ <dim>256</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1918" name="bottleneck4_8/dim_inc/conv/fq_weights_1/zero_point1224522914" type="Const" version="opset1">
+ <data element_type="f16" offset="494664" shape="256,1,1,1" size="512"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1919" name="bottleneck4_8/dim_inc/conv/fq_weights_1/minus_zp" type="Subtract" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>256</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="1">
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>256</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1920" name="bottleneck4_8/dim_inc/conv/fq_weights_1/scale1224020391" type="Const" version="opset1">
+ <data element_type="f16" offset="495176" shape="256,1,1,1" size="512"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1921" name="bottleneck4_8/dim_inc/conv/fq_weights_1/mulpiply_by_scale" type="Multiply" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>256</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="1">
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>256</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1922" name="bottleneck4_8/dim_inc/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1,1" pads_begin="0,0" pads_end="0,0" strides="1,1"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1">
+ <dim>256</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1923" name="data_add_2446524470139319638" type="Const" version="opset1">
+ <data element_type="f16" offset="495688" shape="1,256,1,1" size="512"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1924" name="bottleneck4_8/dim_inc/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_8/dim_inc/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1925" name="bottleneck4_8/add/fq_input_1" type="FakeQuantize" version="opset1">
+ <data auto_broadcast="numpy" levels="256"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1"/>
+ <port id="2"/>
+ <port id="3"/>
+ <port id="4"/>
+ </input>
+ <output>
+ <port id="5" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1926" name="bottleneck4_8/add" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_8/add" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1927" name="bottleneck4_8/fn/weights30920401211396" type="Const" version="opset1">
+ <data element_type="f32" offset="1576" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1928" name="bottleneck4_8/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_8/add" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1929" name="bottleneck4_9/add/fq_input_0" type="FakeQuantize" version="opset1">
+ <data auto_broadcast="numpy" levels="256"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1"/>
+ <port id="2"/>
+ <port id="3"/>
+ <port id="4"/>
+ </input>
+ <output>
+ <port id="5" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1930" name="2974297822422" type="Const" version="opset1">
+ <data element_type="f16" offset="496200" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="1931" name="2975297920355" type="Const" version="opset1">
+ <data element_type="f16" offset="496202" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="1932" name="2976298020496" type="Const" version="opset1">
+ <data element_type="f16" offset="496200" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="1933" name="2977298120955" type="Const" version="opset1">
+ <data element_type="f16" offset="496202" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="1934" name="4904490822641" type="Const" version="opset1">
+ <data element_type="f16" offset="496204" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="1935" name="4905490922407" type="Const" version="opset1">
+ <data element_type="f16" offset="496206" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="1936" name="4906491019527" type="Const" version="opset1">
+ <data element_type="f16" offset="496204" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="1937" name="4907491122473" type="Const" version="opset1">
+ <data element_type="f16" offset="496206" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="1938" name="4984498821795" type="Const" version="opset1">
+ <data element_type="f16" offset="496208" shape="1,64,1,1" size="128"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1939" name="4985498922278" type="Const" version="opset1">
+ <data element_type="f16" offset="496336" shape="1,64,1,1" size="128"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1940" name="4986499021240" type="Const" version="opset1">
+ <data element_type="f16" offset="496208" shape="1,64,1,1" size="128"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1941" name="4987499119680" type="Const" version="opset1">
+ <data element_type="f16" offset="496336" shape="1,64,1,1" size="128"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1942" name="bottleneck4_9/dim_red/bn/mean/Fused_Mul__copy139810385/quantized1216020244" type="Const" version="opset1">
+ <data element_type="i8" offset="496464" shape="64,256,1,1" size="16384"/>
+ <output>
+ <port id="0" precision="I8">
+ <dim>64</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1943" name="bottleneck4_9/dim_red/bn/mean/Fused_Mul__copy139810385/quantized/to_f16" type="Convert" version="opset1">
+ <data destination_type="f16"/>
+ <input>
+ <port id="0">
+ <dim>64</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="1" precision="FP16">
+ <dim>64</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1944" name="bottleneck4_9/dim_red/conv/fq_weights_1/zero_point1217319644" type="Const" version="opset1">
+ <data element_type="f16" offset="512848" shape="64,1,1,1" size="128"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1945" name="bottleneck4_9/dim_red/conv/fq_weights_1/minus_zp" type="Subtract" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>64</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="1">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>64</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1946" name="bottleneck4_9/dim_red/conv/fq_weights_1/scale1216822608" type="Const" version="opset1">
+ <data element_type="f16" offset="512976" shape="64,1,1,1" size="128"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1947" name="bottleneck4_9/dim_red/conv/fq_weights_1/mulpiply_by_scale" type="Multiply" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>64</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="1">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>64</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1948" name="bottleneck4_9/dim_red/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1,1" pads_begin="0,0" pads_end="0,0" strides="1,1"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1">
+ <dim>64</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1949" name="data_add_2447324478140021315" type="Const" version="opset1">
+ <data element_type="f16" offset="513104" shape="1,64,1,1" size="128"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1950" name="bottleneck4_9/dim_red/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_9/dim_red/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1951" name="bottleneck4_9/dim_red/fn/weights30936397281402" type="Const" version="opset1">
+ <data element_type="f32" offset="1576" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1952" name="bottleneck4_9/dim_red/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_9/dim_red/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1953" name="bottleneck4_9/inner/dw1/conv/fq_input_0" type="FakeQuantize" version="opset1">
+ <data auto_broadcast="numpy" levels="256"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="2">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="3">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="4">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="5" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1954" name="16915/value1691719677" type="Const" version="opset1">
+ <data element_type="i64" offset="189984" shape="5" size="40"/>
+ <output>
+ <port id="0" precision="I64">
+ <dim>5</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1955" name="bottleneck4_9/inner/dw1/bn/mean/Fused_Mul__copy140410388/quantized1228020043" type="Const" version="opset1">
+ <data element_type="i8" offset="513232" shape="64,1,3,3" size="576"/>
+ <output>
+ <port id="0" precision="I8">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1956" name="bottleneck4_9/inner/dw1/bn/mean/Fused_Mul__copy140410388/quantized/to_f16" type="Convert" version="opset1">
+ <data destination_type="f16"/>
+ <input>
+ <port id="0">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </input>
+ <output>
+ <port id="1" precision="FP16">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1957" name="bottleneck4_9/inner/dw1/conv/fq_weights_1/zero_point1229319707" type="Const" version="opset1">
+ <data element_type="f16" offset="513808" shape="64,1,1,1" size="128"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1958" name="bottleneck4_9/inner/dw1/conv/fq_weights_1/minus_zp" type="Subtract" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ <port id="1">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1959" name="bottleneck4_9/inner/dw1/conv/fq_weights_1/scale1228820058" type="Const" version="opset1">
+ <data element_type="f16" offset="513936" shape="64,1,1,1" size="128"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1960" name="bottleneck4_9/inner/dw1/conv/fq_weights_1/mulpiply_by_scale" type="Multiply" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ <port id="1">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1961" name="16915" type="Reshape" version="opset1">
+ <data special_zero="true"/>
+ <input>
+ <port id="0">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ <port id="1">
+ <dim>5</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1962" name="bottleneck4_9/inner/dw1/conv" type="GroupConvolution" version="opset1">
+ <data auto_pad="explicit" dilations="1,1" pads_begin="1,1" pads_end="1,1" strides="1,1"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1963" name="data_add_2448124486140620325" type="Const" version="opset1">
+ <data element_type="f16" offset="514064" shape="1,64,1,1" size="128"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1964" name="bottleneck4_9/inner/dw1/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_9/inner/dw1/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1965" name="bottleneck4_9/inner/dw1/fn/weights31000398631408" type="Const" version="opset1">
+ <data element_type="f32" offset="1576" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1966" name="bottleneck4_9/inner/dw1/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_9/inner/dw1/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1967" name="bottleneck4_9/dim_inc/conv/fq_input_0" type="FakeQuantize" version="opset1">
+ <data auto_broadcast="numpy" levels="256"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1"/>
+ <port id="2"/>
+ <port id="3"/>
+ <port id="4"/>
+ </input>
+ <output>
+ <port id="5" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1968" name="bottleneck4_9/dim_inc/bn/mean/Fused_Mul__copy141010391/quantized1304820226" type="Const" version="opset1">
+ <data element_type="i8" offset="514192" shape="256,64,1,1" size="16384"/>
+ <output>
+ <port id="0" precision="I8">
+ <dim>256</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1969" name="bottleneck4_9/dim_inc/bn/mean/Fused_Mul__copy141010391/quantized/to_f16" type="Convert" version="opset1">
+ <data destination_type="f16"/>
+ <input>
+ <port id="0">
+ <dim>256</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="1" precision="FP16">
+ <dim>256</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1970" name="bottleneck4_9/dim_inc/conv/fq_weights_1/zero_point1306120388" type="Const" version="opset1">
+ <data element_type="f16" offset="530576" shape="256,1,1,1" size="512"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1971" name="bottleneck4_9/dim_inc/conv/fq_weights_1/minus_zp" type="Subtract" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>256</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="1">
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>256</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1972" name="bottleneck4_9/dim_inc/conv/fq_weights_1/scale1305622743" type="Const" version="opset1">
+ <data element_type="f16" offset="531088" shape="256,1,1,1" size="512"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1973" name="bottleneck4_9/dim_inc/conv/fq_weights_1/mulpiply_by_scale" type="Multiply" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>256</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="1">
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>256</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1974" name="bottleneck4_9/dim_inc/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1,1" pads_begin="0,0" pads_end="0,0" strides="1,1"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1">
+ <dim>256</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1975" name="data_add_2448924494141220499" type="Const" version="opset1">
+ <data element_type="f16" offset="531600" shape="1,256,1,1" size="512"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1976" name="bottleneck4_9/dim_inc/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_9/dim_inc/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1977" name="bottleneck4_9/add/fq_input_1" type="FakeQuantize" version="opset1">
+ <data auto_broadcast="numpy" levels="256"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1"/>
+ <port id="2"/>
+ <port id="3"/>
+ <port id="4"/>
+ </input>
+ <output>
+ <port id="5" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1978" name="bottleneck4_9/add" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_9/add" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1979" name="bottleneck4_9/fn/weights30924401331415" type="Const" version="opset1">
+ <data element_type="f32" offset="1576" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1980" name="bottleneck4_9/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_9/add" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1981" name="bottleneck4_10/add/fq_input_0" type="FakeQuantize" version="opset1">
+ <data auto_broadcast="numpy" levels="256"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1"/>
+ <port id="2"/>
+ <port id="3"/>
+ <port id="4"/>
+ </input>
+ <output>
+ <port id="5" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1982" name="5454545819371" type="Const" version="opset1">
+ <data element_type="f16" offset="532112" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="1983" name="5455545922059" type="Const" version="opset1">
+ <data element_type="f16" offset="532114" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="1984" name="5456546021150" type="Const" version="opset1">
+ <data element_type="f16" offset="532112" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="1985" name="5457546121081" type="Const" version="opset1">
+ <data element_type="f16" offset="532114" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="1986" name="5124512821459" type="Const" version="opset1">
+ <data element_type="f16" offset="532116" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="1987" name="5125512921867" type="Const" version="opset1">
+ <data element_type="f16" offset="532118" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="1988" name="5126513020775" type="Const" version="opset1">
+ <data element_type="f16" offset="532116" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="1989" name="5127513120931" type="Const" version="opset1">
+ <data element_type="f16" offset="532118" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="1990" name="3484348822092" type="Const" version="opset1">
+ <data element_type="f16" offset="532120" shape="1,64,1,1" size="128"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1991" name="3485348921060" type="Const" version="opset1">
+ <data element_type="f16" offset="532248" shape="1,64,1,1" size="128"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1992" name="3486349022581" type="Const" version="opset1">
+ <data element_type="f16" offset="532120" shape="1,64,1,1" size="128"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1993" name="3487349122227" type="Const" version="opset1">
+ <data element_type="f16" offset="532248" shape="1,64,1,1" size="128"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1994" name="bottleneck4_10/dim_red/bn/mean/Fused_Mul__copy141710394/quantized1187219908" type="Const" version="opset1">
+ <data element_type="i8" offset="532376" shape="64,256,1,1" size="16384"/>
+ <output>
+ <port id="0" precision="I8">
+ <dim>64</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1995" name="bottleneck4_10/dim_red/bn/mean/Fused_Mul__copy141710394/quantized/to_f16" type="Convert" version="opset1">
+ <data destination_type="f16"/>
+ <input>
+ <port id="0">
+ <dim>64</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="1" precision="FP16">
+ <dim>64</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1996" name="bottleneck4_10/dim_red/conv/fq_weights_1/zero_point1188522767" type="Const" version="opset1">
+ <data element_type="f16" offset="548760" shape="64,1,1,1" size="128"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1997" name="bottleneck4_10/dim_red/conv/fq_weights_1/minus_zp" type="Subtract" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>64</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="1">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>64</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1998" name="bottleneck4_10/dim_red/conv/fq_weights_1/scale1188021270" type="Const" version="opset1">
+ <data element_type="f16" offset="548888" shape="64,1,1,1" size="128"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1999" name="bottleneck4_10/dim_red/conv/fq_weights_1/mulpiply_by_scale" type="Multiply" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>64</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="1">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>64</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="2000" name="bottleneck4_10/dim_red/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1,1" pads_begin="0,0" pads_end="0,0" strides="1,1"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1">
+ <dim>64</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="2001" name="data_add_2449724502141920214" type="Const" version="opset1">
+ <data element_type="f16" offset="549016" shape="1,64,1,1" size="128"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="2002" name="bottleneck4_10/dim_red/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_10/dim_red/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="2003" name="bottleneck4_10/dim_red/fn/weights31168400491421" type="Const" version="opset1">
+ <data element_type="f32" offset="1576" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="2004" name="bottleneck4_10/dim_red/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_10/dim_red/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="2005" name="bottleneck4_10/inner/dw1/conv/fq_input_0" type="FakeQuantize" version="opset1">
+ <data auto_broadcast="numpy" levels="256"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="2">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="3">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="4">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="5" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="2006" name="16839/value1684122488" type="Const" version="opset1">
+ <data element_type="i64" offset="189984" shape="5" size="40"/>
+ <output>
+ <port id="0" precision="I64">
+ <dim>5</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="2007" name="bottleneck4_10/inner/dw1/bn/mean/Fused_Mul__copy142310397/quantized1357622866" type="Const" version="opset1">
+ <data element_type="i8" offset="549144" shape="64,1,3,3" size="576"/>
+ <output>
+ <port id="0" precision="I8">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="2008" name="bottleneck4_10/inner/dw1/bn/mean/Fused_Mul__copy142310397/quantized/to_f16" type="Convert" version="opset1">
+ <data destination_type="f16"/>
+ <input>
+ <port id="0">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </input>
+ <output>
+ <port id="1" precision="FP16">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="2009" name="bottleneck4_10/inner/dw1/conv/fq_weights_1/zero_point1358922440" type="Const" version="opset1">
+ <data element_type="f16" offset="549720" shape="64,1,1,1" size="128"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="2010" name="bottleneck4_10/inner/dw1/conv/fq_weights_1/minus_zp" type="Subtract" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ <port id="1">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="2011" name="bottleneck4_10/inner/dw1/conv/fq_weights_1/scale1358420976" type="Const" version="opset1">
+ <data element_type="f16" offset="549848" shape="64,1,1,1" size="128"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="2012" name="bottleneck4_10/inner/dw1/conv/fq_weights_1/mulpiply_by_scale" type="Multiply" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ <port id="1">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="2013" name="16839" type="Reshape" version="opset1">
+ <data special_zero="true"/>
+ <input>
+ <port id="0">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ <port id="1">
+ <dim>5</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="2014" name="bottleneck4_10/inner/dw1/conv" type="GroupConvolution" version="opset1">
+ <data auto_pad="explicit" dilations="1,1" pads_begin="1,1" pads_end="1,1" strides="1,1"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="2015" name="data_add_2450524510142519743" type="Const" version="opset1">
+ <data element_type="f16" offset="549976" shape="1,64,1,1" size="128"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="2016" name="bottleneck4_10/inner/dw1/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_10/inner/dw1/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="2017" name="bottleneck4_10/inner/dw1/fn/weights30944402621427" type="Const" version="opset1">
+ <data element_type="f32" offset="1576" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="2018" name="bottleneck4_10/inner/dw1/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_10/inner/dw1/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="2019" name="bottleneck4_10/dim_inc/conv/fq_input_0" type="FakeQuantize" version="opset1">
+ <data auto_broadcast="numpy" levels="256"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1"/>
+ <port id="2"/>
+ <port id="3"/>
+ <port id="4"/>
+ </input>
+ <output>
+ <port id="5" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="2020" name="bottleneck4_10/dim_inc/bn/mean/Fused_Mul__copy142910400/quantized1364819974" type="Const" version="opset1">
+ <data element_type="i8" offset="550104" shape="256,64,1,1" size="16384"/>
+ <output>
+ <port id="0" precision="I8">
+ <dim>256</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="2021" name="bottleneck4_10/dim_inc/bn/mean/Fused_Mul__copy142910400/quantized/to_f16" type="Convert" version="opset1">
+ <data destination_type="f16"/>
+ <input>
+ <port id="0">
+ <dim>256</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="1" precision="FP16">
+ <dim>256</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="2022" name="bottleneck4_10/dim_inc/conv/fq_weights_1/zero_point1366122704" type="Const" version="opset1">
+ <data element_type="f16" offset="566488" shape="256,1,1,1" size="512"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="2023" name="bottleneck4_10/dim_inc/conv/fq_weights_1/minus_zp" type="Subtract" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>256</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="1">
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>256</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="2024" name="bottleneck4_10/dim_inc/conv/fq_weights_1/scale1365622392" type="Const" version="opset1">
+ <data element_type="f16" offset="567000" shape="256,1,1,1" size="512"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="2025" name="bottleneck4_10/dim_inc/conv/fq_weights_1/mulpiply_by_scale" type="Multiply" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>256</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="1">
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>256</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="2026" name="bottleneck4_10/dim_inc/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1,1" pads_begin="0,0" pads_end="0,0" strides="1,1"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1">
+ <dim>256</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="2027" name="data_add_2451324518143120667" type="Const" version="opset1">
+ <data element_type="f16" offset="567512" shape="1,256,1,1" size="512"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="2028" name="bottleneck4_10/dim_inc/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_10/dim_inc/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="2029" name="bottleneck4_10/add/fq_input_1" type="FakeQuantize" version="opset1">
+ <data auto_broadcast="numpy" levels="256"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1"/>
+ <port id="2"/>
+ <port id="3"/>
+ <port id="4"/>
+ </input>
+ <output>
+ <port id="5" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="2030" name="bottleneck4_10/add" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_10/add" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="2031" name="bottleneck4_10/fn/weights30904403071434" type="Const" version="opset1">
+ <data element_type="f32" offset="1576" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="2032" name="bottleneck4_10/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_10/add" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="2033" name="bottleneck4_11/add/fq_input_0" type="FakeQuantize" version="opset1">
+ <data auto_broadcast="numpy" levels="256"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1"/>
+ <port id="2"/>
+ <port id="3"/>
+ <port id="4"/>
+ </input>
+ <output>
+ <port id="5" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="2034" name="4294429819989" type="Const" version="opset1">
+ <data element_type="f16" offset="568024" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="2035" name="4295429920127" type="Const" version="opset1">
+ <data element_type="f16" offset="568026" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="2036" name="4296430021717" type="Const" version="opset1">
+ <data element_type="f16" offset="568024" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="2037" name="4297430121204" type="Const" version="opset1">
+ <data element_type="f16" offset="568026" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="2038" name="5324532822170" type="Const" version="opset1">
+ <data element_type="f16" offset="568028" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="2039" name="5325532922719" type="Const" version="opset1">
+ <data element_type="f16" offset="568030" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="2040" name="5326533020073" type="Const" version="opset1">
+ <data element_type="f16" offset="568028" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="2041" name="5327533122332" type="Const" version="opset1">
+ <data element_type="f16" offset="568030" shape="" size="2"/>
+ <output>
+ <port id="0" precision="FP16"/>
+ </output>
+ </layer>
+ <layer id="2042" name="3304330820979" type="Const" version="opset1">
+ <data element_type="f16" offset="568032" shape="1,64,1,1" size="128"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="2043" name="3305330922551" type="Const" version="opset1">
+ <data element_type="f16" offset="568160" shape="1,64,1,1" size="128"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="2044" name="3306331022683" type="Const" version="opset1">
+ <data element_type="f16" offset="568032" shape="1,64,1,1" size="128"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="2045" name="3307331120856" type="Const" version="opset1">
+ <data element_type="f16" offset="568160" shape="1,64,1,1" size="128"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="2046" name="bottleneck4_11/dim_red/bn/mean/Fused_Mul__copy143610403/quantized1362419785" type="Const" version="opset1">
+ <data element_type="i8" offset="568288" shape="64,256,1,1" size="16384"/>
+ <output>
+ <port id="0" precision="I8">
+ <dim>64</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="2047" name="bottleneck4_11/dim_red/bn/mean/Fused_Mul__copy143610403/quantized/to_f16" type="Convert" version="opset1">
+ <data destination_type="f16"/>
+ <input>
+ <port id="0">
+ <dim>64</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="1" precision="FP16">
+ <dim>64</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="2048" name="bottleneck4_11/dim_red/conv/fq_weights_1/zero_point1363720406" type="Const" version="opset1">
+ <data element_type="f16" offset="584672" shape="64,1,1,1" size="128"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="2049" name="bottleneck4_11/dim_red/conv/fq_weights_1/minus_zp" type="Subtract" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>64</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="1">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>64</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="2050" name="bottleneck4_11/dim_red/conv/fq_weights_1/scale1363222524" type="Const" version="opset1">
+ <data element_type="f16" offset="584800" shape="64,1,1,1" size="128"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="2051" name="bottleneck4_11/dim_red/conv/fq_weights_1/mulpiply_by_scale" type="Multiply" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>64</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="1">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>64</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="2052" name="bottleneck4_11/dim_red/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1,1" pads_begin="0,0" pads_end="0,0" strides="1,1"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1">
+ <dim>64</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="2053" name="data_add_2452124526143821531" type="Const" version="opset1">
+ <data element_type="f16" offset="584928" shape="1,64,1,1" size="128"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="2054" name="bottleneck4_11/dim_red/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_11/dim_red/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="2055" name="bottleneck4_11/dim_red/fn/weights30956405321440" type="Const" version="opset1">
+ <data element_type="f32" offset="1576" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="2056" name="bottleneck4_11/dim_red/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_11/dim_red/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="2057" name="bottleneck4_11/inner/dw1/conv/fq_input_0" type="FakeQuantize" version="opset1">
+ <data auto_broadcast="numpy" levels="256"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="2">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="3">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="4">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="5" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="2058" name="16827/value1682922857" type="Const" version="opset1">
+ <data element_type="i64" offset="189984" shape="5" size="40"/>
+ <output>
+ <port id="0" precision="I64">
+ <dim>5</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="2059" name="bottleneck4_11/inner/dw1/bn/mean/Fused_Mul__copy144210406/quantized1292821633" type="Const" version="opset1">
+ <data element_type="i8" offset="585056" shape="64,1,3,3" size="576"/>
+ <output>
+ <port id="0" precision="I8">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="2060" name="bottleneck4_11/inner/dw1/bn/mean/Fused_Mul__copy144210406/quantized/to_f16" type="Convert" version="opset1">
+ <data destination_type="f16"/>
+ <input>
+ <port id="0">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </input>
+ <output>
+ <port id="1" precision="FP16">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="2061" name="bottleneck4_11/inner/dw1/conv/fq_weights_1/zero_point1294120229" type="Const" version="opset1">
+ <data element_type="f16" offset="585632" shape="64,1,1,1" size="128"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="2062" name="bottleneck4_11/inner/dw1/conv/fq_weights_1/minus_zp" type="Subtract" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ <port id="1">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="2063" name="bottleneck4_11/inner/dw1/conv/fq_weights_1/scale1293620895" type="Const" version="opset1">
+ <data element_type="f16" offset="585760" shape="64,1,1,1" size="128"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="2064" name="bottleneck4_11/inner/dw1/conv/fq_weights_1/mulpiply_by_scale" type="Multiply" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ <port id="1">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="2065" name="16827" type="Reshape" version="opset1">
+ <data special_zero="true"/>
+ <input>
+ <port id="0">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ <port id="1">
+ <dim>5</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="2066" name="bottleneck4_11/inner/dw1/conv" type="GroupConvolution" version="opset1">
+ <data auto_pad="explicit" dilations="1,1" pads_begin="1,1" pads_end="1,1" strides="1,1"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="2067" name="data_add_2452924534144422578" type="Const" version="opset1">
+ <data element_type="f16" offset="585888" shape="1,64,1,1" size="128"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="2068" name="bottleneck4_11/inner/dw1/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_11/inner/dw1/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="2069" name="bottleneck4_11/inner/dw1/fn/weights30776403011446" type="Const" version="opset1">
+ <data element_type="f32" offset="1576" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="2070" name="bottleneck4_11/inner/dw1/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_11/inner/dw1/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="2071" name="bottleneck4_11/dim_inc/conv/fq_input_0" type="FakeQuantize" version="opset1">
+ <data auto_broadcast="numpy" levels="256"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1"/>
+ <port id="2"/>
+ <port id="3"/>
+ <port id="4"/>
+ </input>
+ <output>
+ <port id="5" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="2072" name="bottleneck4_11/dim_inc/bn/mean/Fused_Mul__copy144810409/quantized1384021810" type="Const" version="opset1">
+ <data element_type="i8" offset="586016" shape="256,64,1,1" size="16384"/>
+ <output>
+ <port id="0" precision="I8">
+ <dim>256</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="2073" name="bottleneck4_11/dim_inc/bn/mean/Fused_Mul__copy144810409/quantized/to_f16" type="Convert" version="opset1">
+ <data destination_type="f16"/>
+ <input>
+ <port id="0">
+ <dim>256</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="1" precision="FP16">
+ <dim>256</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="2074" name="bottleneck4_11/dim_inc/conv/fq_weights_1/zero_point1385319533" type="Const" version="opset1">
+ <data element_type="f16" offset="602400" shape="256,1,1,1" size="512"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="2075" name="bottleneck4_11/dim_inc/conv/fq_weights_1/minus_zp" type="Subtract" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>256</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="1">
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>256</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="2076" name="bottleneck4_11/dim_inc/conv/fq_weights_1/scale1384821798" type="Const" version="opset1">
+ <data element_type="f16" offset="602912" shape="256,1,1,1" size="512"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="2077" name="bottleneck4_11/dim_inc/conv/fq_weights_1/mulpiply_by_scale" type="Multiply" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>256</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ <port id="1">
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>256</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="2078" name="bottleneck4_11/dim_inc/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1,1" pads_begin="0,0" pads_end="0,0" strides="1,1"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1">
+ <dim>256</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="2079" name="data_add_2453724542145022566" type="Const" version="opset1">
+ <data element_type="f16" offset="603424" shape="1,256,1,1" size="512"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="2080" name="bottleneck4_11/dim_inc/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_11/dim_inc/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="2081" name="bottleneck4_11/add/fq_input_1" type="FakeQuantize" version="opset1">
+ <data auto_broadcast="numpy" levels="256"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1"/>
+ <port id="2"/>
+ <port id="3"/>
+ <port id="4"/>
+ </input>
+ <output>
+ <port id="5" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="2082" name="bottleneck4_11/add" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_11/add" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="2083" name="bottleneck4_11/fn/weights31180405651453" type="Const" version="opset1">
+ <data element_type="f32" offset="1576" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="2084" name="bottleneck4_11/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bb_16xout_pd" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="2085" name="mbox_conf1/out/conv/WithoutBiases/fq_input_0" type="FakeQuantize" version="opset1">
+ <data auto_broadcast="numpy" levels="256"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1"/>
+ <port id="2"/>
+ <port id="3"/>
+ <port id="4"/>
+ </input>
+ <output>
+ <port id="5" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="2086" name="752145510412/quantized1300022731" type="Const" version="opset1">
+ <data element_type="i8" offset="603936" shape="48,256,3,3" size="110592"/>
+ <output>
+ <port id="0" precision="I8">
+ <dim>48</dim>
+ <dim>256</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="2087" name="752145510412/quantized/to_f16" type="Convert" version="opset1">
+ <data destination_type="f16"/>
+ <input>
+ <port id="0">
+ <dim>48</dim>
+ <dim>256</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </input>
+ <output>
+ <port id="1" precision="FP16">
+ <dim>48</dim>
+ <dim>256</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="2088" name="mbox_loc1/out/conv/WithoutBiases/fq_weights_1/zero_point1301321714" type="Const" version="opset1">
+ <data element_type="f16" offset="714528" shape="48,1,1,1" size="96"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>48</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="2089" name="mbox_loc1/out/conv/WithoutBiases/fq_weights_1/minus_zp" type="Subtract" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>48</dim>
+ <dim>256</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ <port id="1">
+ <dim>48</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>48</dim>
+ <dim>256</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="2090" name="mbox_loc1/out/conv/WithoutBiases/fq_weights_1/scale1300820010" type="Const" version="opset1">
+ <data element_type="f16" offset="714624" shape="48,1,1,1" size="96"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>48</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="2091" name="mbox_loc1/out/conv/WithoutBiases/fq_weights_1/mulpiply_by_scale" type="Multiply" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>48</dim>
+ <dim>256</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ <port id="1">
+ <dim>48</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>48</dim>
+ <dim>256</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="2092" name="mbox_loc1/out/conv/WithoutBiases" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1,1" pads_begin="1,1" pads_end="1,1" strides="1,1"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1">
+ <dim>48</dim>
+ <dim>256</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>48</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="2093" name="mbox_loc1/out/conv/Dims13831145721933" type="Const" version="opset1">
+ <data element_type="f16" offset="714720" shape="1,48,1,1" size="96"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>48</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="2094" name="mbox_loc1/out/conv" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>48</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ <dim>48</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="mbox_loc1/out/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>48</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="2095" name="12961459" type="Const" version="opset1">
+ <data element_type="i64" offset="714816" shape="4" size="32"/>
+ <output>
+ <port id="0" precision="I64">
+ <dim>4</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="2096" name="mbox_loc1/out/conv/perm" type="Transpose" version="opset1">
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>48</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1">
+ <dim>4</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="mbox_loc1/out/conv/perm" precision="FP16">
+ <dim>1</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ <dim>48</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="2097" name="1303/shapes_concat146120100" type="Const" version="opset1">
+ <data element_type="i64" offset="714848" shape="2" size="16"/>
+ <output>
+ <port id="0" precision="I64">
+ <dim>2</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="2098" name="mbox_loc1/out/conv/flat" type="Reshape" version="opset1">
+ <data special_zero="true"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ <dim>48</dim>
+ </port>
+ <port id="1">
+ <dim>2</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="mbox_loc1/out/conv/flat" precision="FP16">
+ <dim>1</dim>
+ <dim>32640</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="2099" name="811146310416/quantized1261621258" type="Const" version="opset1">
+ <data element_type="i8" offset="714864" shape="24,256,3,3" size="55296"/>
+ <output>
+ <port id="0" precision="I8">
+ <dim>24</dim>
+ <dim>256</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="2100" name="811146310416/quantized/to_f16" type="Convert" version="opset1">
+ <data destination_type="f16"/>
+ <input>
+ <port id="0">
+ <dim>24</dim>
+ <dim>256</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </input>
+ <output>
+ <port id="1" precision="FP16">
+ <dim>24</dim>
+ <dim>256</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="2101" name="mbox_conf1/out/conv/WithoutBiases/fq_weights_1/zero_point1262920310" type="Const" version="opset1">
+ <data element_type="f16" offset="770160" shape="24,1,1,1" size="48"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>24</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="2102" name="mbox_conf1/out/conv/WithoutBiases/fq_weights_1/minus_zp" type="Subtract" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>24</dim>
+ <dim>256</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ <port id="1">
+ <dim>24</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>24</dim>
+ <dim>256</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="2103" name="mbox_conf1/out/conv/WithoutBiases/fq_weights_1/scale1262419446" type="Const" version="opset1">
+ <data element_type="f16" offset="770208" shape="24,1,1,1" size="48"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>24</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="2104" name="mbox_conf1/out/conv/WithoutBiases/fq_weights_1/mulpiply_by_scale" type="Multiply" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>24</dim>
+ <dim>256</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ <port id="1">
+ <dim>24</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>24</dim>
+ <dim>256</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="2105" name="mbox_conf1/out/conv/WithoutBiases" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1,1" pads_begin="1,1" pads_end="1,1" strides="1,1"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1">
+ <dim>24</dim>
+ <dim>256</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>24</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="2106" name="mbox_conf1/out/conv/Dims13825146521111" type="Const" version="opset1">
+ <data element_type="f16" offset="770256" shape="1,24,1,1" size="48"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>24</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="2107" name="mbox_conf1/out/conv" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>24</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ <dim>24</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="mbox_conf1/out/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>24</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="2108" name="12971467" type="Const" version="opset1">
+ <data element_type="i64" offset="714816" shape="4" size="32"/>
+ <output>
+ <port id="0" precision="I64">
+ <dim>4</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="2109" name="mbox_conf1/out/conv/perm" type="Transpose" version="opset1">
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>24</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1">
+ <dim>4</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="mbox_conf1/out/conv/perm" precision="FP16">
+ <dim>1</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ <dim>24</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="2110" name="1308/shapes_concat146920376" type="Const" version="opset1">
+ <data element_type="i64" offset="714848" shape="2" size="16"/>
+ <output>
+ <port id="0" precision="I64">
+ <dim>2</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="2111" name="mbox_conf1/out/conv/flat" type="Reshape" version="opset1">
+ <data special_zero="true"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ <dim>24</dim>
+ </port>
+ <port id="1">
+ <dim>2</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="mbox_conf1/out/conv/flat" precision="FP16">
+ <dim>1</dim>
+ <dim>16320</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="2112" name="1295147119821" type="Const" version="opset1">
+ <data element_type="i64" offset="770304" shape="3" size="24"/>
+ <output>
+ <port id="0" precision="I64">
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="2113" name="mbox_conf1/out/conv/flat/reshape" type="Reshape" version="opset1">
+ <data special_zero="true"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>16320</dim>
+ </port>
+ <port id="1">
+ <dim>3</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="mbox_conf1/out/conv/flat/reshape" precision="FP16">
+ <dim>1</dim>
+ <dim>8160</dim>
+ <dim>2</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="2114" name="mbox_conf1/out/conv/flat/softmax" type="SoftMax" version="opset1">
+ <data axis="2"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>8160</dim>
+ <dim>2</dim>
+ </port>
+ </input>
+ <output>
+ <port id="1" names="mbox_conf1/out/conv/flat/softmax" precision="FP16">
+ <dim>1</dim>
+ <dim>8160</dim>
+ <dim>2</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="2115" name="1298/shapes_concat147422509" type="Const" version="opset1">
+ <data element_type="i64" offset="714848" shape="2" size="16"/>
+ <output>
+ <port id="0" precision="I64">
+ <dim>2</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="2116" name="mbox_conf1/out/conv/flat/softmax/flat" type="Reshape" version="opset1">
+ <data special_zero="true"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>8160</dim>
+ <dim>2</dim>
+ </port>
+ <port id="1">
+ <dim>2</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="mbox_conf1/out/conv/flat/softmax/flat" precision="FP16">
+ <dim>1</dim>
+ <dim>16320</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="2117" name="mbox1/priorbox/0_port" type="ShapeOf" version="opset3">
+ <data output_type="i64"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </input>
+ <output>
+ <port id="1" precision="I64">
+ <dim>4</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="2118" name="mbox1/priorbox/ss_begin29786405591477" type="Const" version="opset1">
+ <data element_type="i64" offset="770328" shape="1" size="8"/>
+ <output>
+ <port id="0" precision="I64">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="2119" name="mbox1/priorbox/ss_end29787403131478" type="Const" version="opset1">
+ <data element_type="i64" offset="770336" shape="1" size="8"/>
+ <output>
+ <port id="0" precision="I64">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="2120" name="mbox1/priorbox/ss_stride29788398871479" type="Const" version="opset1">
+ <data element_type="i64" offset="770344" shape="1" size="8"/>
+ <output>
+ <port id="0" precision="I64">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="2121" name="mbox1/priorbox/ss_0_port" type="StridedSlice" version="opset1">
+ <data begin_mask="0" ellipsis_mask="0" end_mask="1" new_axis_mask="0" shrink_axis_mask="0"/>
+ <input>
+ <port id="0">
+ <dim>4</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ </port>
+ <port id="2">
+ <dim>1</dim>
+ </port>
+ <port id="3">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="4" precision="I64">
+ <dim>2</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="2122" name="mbox1/priorbox/1_port" type="ShapeOf" version="opset3">
+ <data output_type="i64"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>320</dim>
+ <dim>544</dim>
+ </port>
+ </input>
+ <output>
+ <port id="1" precision="I64">
+ <dim>4</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="2123" name="mbox1/priorbox/ss_begin29786405861482" type="Const" version="opset1">
+ <data element_type="i64" offset="770328" shape="1" size="8"/>
+ <output>
+ <port id="0" precision="I64">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="2124" name="mbox1/priorbox/ss_end29787398301483" type="Const" version="opset1">
+ <data element_type="i64" offset="770336" shape="1" size="8"/>
+ <output>
+ <port id="0" precision="I64">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="2125" name="mbox1/priorbox/ss_stride29788405201484" type="Const" version="opset1">
+ <data element_type="i64" offset="770344" shape="1" size="8"/>
+ <output>
+ <port id="0" precision="I64">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="2126" name="mbox1/priorbox/ss_1_port" type="StridedSlice" version="opset1">
+ <data begin_mask="0" ellipsis_mask="0" end_mask="1" new_axis_mask="0" shrink_axis_mask="0"/>
+ <input>
+ <port id="0">
+ <dim>4</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ </port>
+ <port id="2">
+ <dim>1</dim>
+ </port>
+ <port id="3">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="4" precision="I64">
+ <dim>2</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="2127" name="mbox1/priorbox/naked_not_unsqueezed" type="PriorBoxClustered" version="opset1">
+ <data clip="0" height="34.07,47.11,54.22,65.78,75.56,80.89,89.78,99.26,115.56,163.26,194.07,197.33" offset="0.5" step="0" step_h="16" step_w="16" variance="0.1,0.1,0.2,0.2" width="11.33,17,20.68,23.52,28.05,37.4,30.03,35.7,44.2,55.25,78.12,135.15"/>
+ <input>
+ <port id="0">
+ <dim>2</dim>
+ </port>
+ <port id="1">
+ <dim>2</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP32">
+ <dim>2</dim>
+ <dim>32640</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="2128" name="mbox1/priorbox/unsqueeze/value2979640538148720766" type="Const" version="opset1">
+ <data element_type="i64" offset="770352" shape="1" size="8"/>
+ <output>
+ <port id="0" precision="I64">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="2129" name="mbox1/priorbox" type="Unsqueeze" version="opset1">
+ <input>
+ <port id="0">
+ <dim>2</dim>
+ <dim>32640</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="mbox1/priorbox" precision="FP32">
+ <dim>1</dim>
+ <dim>2</dim>
+ <dim>32640</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="2130" name="detection_out" type="DetectionOutput" version="opset1">
+ <data background_label_id="0" clip_after_nms="false" clip_before_nms="false" code_type="caffe.PriorBoxParameter.CENTER_SIZE" confidence_threshold="0.009999999776482582" decrease_label_id="false" input_height="1" input_width="1" keep_top_k="200" nms_threshold="0.44999998807907104" normalized="true" num_classes="2" objectness_score="0" share_location="true" top_k="400" variance_encoded_in_target="false"/>
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>32640</dim>
+ </port>
+ <port id="1">
+ <dim>1</dim>
+ <dim>16320</dim>
+ </port>
+ <port id="2">
+ <dim>1</dim>
+ <dim>2</dim>
+ <dim>32640</dim>
+ </port>
+ </input>
+ <output>
+ <port id="3" names="detection_out" precision="FP32">
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>200</dim>
+ <dim>7</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="2131" name="detection_out/sink_port_0" type="Result" version="opset1">
+ <input>
+ <port id="0">
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>200</dim>
+ <dim>7</dim>
+ </port>
+ </input>
+ </layer>
+ </layers>
+ <edges>
+ <edge from-layer="164" from-port="0" to-layer="166" to-port="0"/>
+ <edge from-layer="165" from-port="0" to-layer="166" to-port="1"/>
+ <edge from-layer="166" from-port="2" to-layer="168" to-port="0"/>
+ <edge from-layer="167" from-port="0" to-layer="168" to-port="1"/>
+ <edge from-layer="168" from-port="2" to-layer="169" to-port="0"/>
+ <edge from-layer="160" from-port="0" to-layer="169" to-port="1"/>
+ <edge from-layer="161" from-port="0" to-layer="169" to-port="2"/>
+ <edge from-layer="162" from-port="0" to-layer="169" to-port="3"/>
+ <edge from-layer="163" from-port="0" to-layer="169" to-port="4"/>
+ <edge from-layer="170" from-port="0" to-layer="171" to-port="0"/>
+ <edge from-layer="171" from-port="1" to-layer="173" to-port="0"/>
+ <edge from-layer="172" from-port="0" to-layer="173" to-port="1"/>
+ <edge from-layer="173" from-port="2" to-layer="175" to-port="0"/>
+ <edge from-layer="174" from-port="0" to-layer="175" to-port="1"/>
+ <edge from-layer="169" from-port="5" to-layer="176" to-port="0"/>
+ <edge from-layer="175" from-port="2" to-layer="176" to-port="1"/>
+ <edge from-layer="176" from-port="2" to-layer="178" to-port="0"/>
+ <edge from-layer="177" from-port="0" to-layer="178" to-port="1"/>
+ <edge from-layer="178" from-port="2" to-layer="179" to-port="0"/>
+ <edge from-layer="179" from-port="1" to-layer="180" to-port="0"/>
+ <edge from-layer="156" from-port="0" to-layer="180" to-port="1"/>
+ <edge from-layer="157" from-port="0" to-layer="180" to-port="2"/>
+ <edge from-layer="158" from-port="0" to-layer="180" to-port="3"/>
+ <edge from-layer="159" from-port="0" to-layer="180" to-port="4"/>
+ <edge from-layer="193" from-port="0" to-layer="194" to-port="0"/>
+ <edge from-layer="194" from-port="1" to-layer="196" to-port="0"/>
+ <edge from-layer="195" from-port="0" to-layer="196" to-port="1"/>
+ <edge from-layer="196" from-port="2" to-layer="198" to-port="0"/>
+ <edge from-layer="197" from-port="0" to-layer="198" to-port="1"/>
+ <edge from-layer="180" from-port="5" to-layer="199" to-port="0"/>
+ <edge from-layer="198" from-port="2" to-layer="199" to-port="1"/>
+ <edge from-layer="199" from-port="2" to-layer="201" to-port="0"/>
+ <edge from-layer="200" from-port="0" to-layer="201" to-port="1"/>
+ <edge from-layer="201" from-port="2" to-layer="203" to-port="0"/>
+ <edge from-layer="202" from-port="0" to-layer="203" to-port="1"/>
+ <edge from-layer="203" from-port="2" to-layer="204" to-port="0"/>
+ <edge from-layer="189" from-port="0" to-layer="204" to-port="1"/>
+ <edge from-layer="190" from-port="0" to-layer="204" to-port="2"/>
+ <edge from-layer="191" from-port="0" to-layer="204" to-port="3"/>
+ <edge from-layer="192" from-port="0" to-layer="204" to-port="4"/>
+ <edge from-layer="206" from-port="0" to-layer="207" to-port="0"/>
+ <edge from-layer="207" from-port="1" to-layer="209" to-port="0"/>
+ <edge from-layer="208" from-port="0" to-layer="209" to-port="1"/>
+ <edge from-layer="209" from-port="2" to-layer="211" to-port="0"/>
+ <edge from-layer="210" from-port="0" to-layer="211" to-port="1"/>
+ <edge from-layer="211" from-port="2" to-layer="212" to-port="0"/>
+ <edge from-layer="205" from-port="0" to-layer="212" to-port="1"/>
+ <edge from-layer="204" from-port="5" to-layer="213" to-port="0"/>
+ <edge from-layer="212" from-port="2" to-layer="213" to-port="1"/>
+ <edge from-layer="213" from-port="2" to-layer="215" to-port="0"/>
+ <edge from-layer="214" from-port="0" to-layer="215" to-port="1"/>
+ <edge from-layer="215" from-port="2" to-layer="217" to-port="0"/>
+ <edge from-layer="216" from-port="0" to-layer="217" to-port="1"/>
+ <edge from-layer="217" from-port="2" to-layer="218" to-port="0"/>
+ <edge from-layer="185" from-port="0" to-layer="218" to-port="1"/>
+ <edge from-layer="186" from-port="0" to-layer="218" to-port="2"/>
+ <edge from-layer="187" from-port="0" to-layer="218" to-port="3"/>
+ <edge from-layer="188" from-port="0" to-layer="218" to-port="4"/>
+ <edge from-layer="219" from-port="0" to-layer="220" to-port="0"/>
+ <edge from-layer="220" from-port="1" to-layer="222" to-port="0"/>
+ <edge from-layer="221" from-port="0" to-layer="222" to-port="1"/>
+ <edge from-layer="222" from-port="2" to-layer="224" to-port="0"/>
+ <edge from-layer="223" from-port="0" to-layer="224" to-port="1"/>
+ <edge from-layer="218" from-port="5" to-layer="225" to-port="0"/>
+ <edge from-layer="224" from-port="2" to-layer="225" to-port="1"/>
+ <edge from-layer="225" from-port="2" to-layer="227" to-port="0"/>
+ <edge from-layer="226" from-port="0" to-layer="227" to-port="1"/>
+ <edge from-layer="227" from-port="2" to-layer="228" to-port="0"/>
+ <edge from-layer="181" from-port="0" to-layer="228" to-port="1"/>
+ <edge from-layer="182" from-port="0" to-layer="228" to-port="2"/>
+ <edge from-layer="183" from-port="0" to-layer="228" to-port="3"/>
+ <edge from-layer="184" from-port="0" to-layer="228" to-port="4"/>
+ <edge from-layer="180" from-port="5" to-layer="229" to-port="0"/>
+ <edge from-layer="228" from-port="5" to-layer="229" to-port="1"/>
+ <edge from-layer="229" from-port="2" to-layer="231" to-port="0"/>
+ <edge from-layer="230" from-port="0" to-layer="231" to-port="1"/>
+ <edge from-layer="231" from-port="2" to-layer="232" to-port="0"/>
+ <edge from-layer="152" from-port="0" to-layer="232" to-port="1"/>
+ <edge from-layer="153" from-port="0" to-layer="232" to-port="2"/>
+ <edge from-layer="154" from-port="0" to-layer="232" to-port="3"/>
+ <edge from-layer="155" from-port="0" to-layer="232" to-port="4"/>
+ <edge from-layer="245" from-port="0" to-layer="246" to-port="0"/>
+ <edge from-layer="246" from-port="1" to-layer="248" to-port="0"/>
+ <edge from-layer="247" from-port="0" to-layer="248" to-port="1"/>
+ <edge from-layer="248" from-port="2" to-layer="250" to-port="0"/>
+ <edge from-layer="249" from-port="0" to-layer="250" to-port="1"/>
+ <edge from-layer="232" from-port="5" to-layer="251" to-port="0"/>
+ <edge from-layer="250" from-port="2" to-layer="251" to-port="1"/>
+ <edge from-layer="251" from-port="2" to-layer="253" to-port="0"/>
+ <edge from-layer="252" from-port="0" to-layer="253" to-port="1"/>
+ <edge from-layer="253" from-port="2" to-layer="255" to-port="0"/>
+ <edge from-layer="254" from-port="0" to-layer="255" to-port="1"/>
+ <edge from-layer="255" from-port="2" to-layer="256" to-port="0"/>
+ <edge from-layer="241" from-port="0" to-layer="256" to-port="1"/>
+ <edge from-layer="242" from-port="0" to-layer="256" to-port="2"/>
+ <edge from-layer="243" from-port="0" to-layer="256" to-port="3"/>
+ <edge from-layer="244" from-port="0" to-layer="256" to-port="4"/>
+ <edge from-layer="258" from-port="0" to-layer="259" to-port="0"/>
+ <edge from-layer="259" from-port="1" to-layer="261" to-port="0"/>
+ <edge from-layer="260" from-port="0" to-layer="261" to-port="1"/>
+ <edge from-layer="261" from-port="2" to-layer="263" to-port="0"/>
+ <edge from-layer="262" from-port="0" to-layer="263" to-port="1"/>
+ <edge from-layer="263" from-port="2" to-layer="264" to-port="0"/>
+ <edge from-layer="257" from-port="0" to-layer="264" to-port="1"/>
+ <edge from-layer="256" from-port="5" to-layer="265" to-port="0"/>
+ <edge from-layer="264" from-port="2" to-layer="265" to-port="1"/>
+ <edge from-layer="265" from-port="2" to-layer="267" to-port="0"/>
+ <edge from-layer="266" from-port="0" to-layer="267" to-port="1"/>
+ <edge from-layer="267" from-port="2" to-layer="269" to-port="0"/>
+ <edge from-layer="268" from-port="0" to-layer="269" to-port="1"/>
+ <edge from-layer="269" from-port="2" to-layer="270" to-port="0"/>
+ <edge from-layer="237" from-port="0" to-layer="270" to-port="1"/>
+ <edge from-layer="238" from-port="0" to-layer="270" to-port="2"/>
+ <edge from-layer="239" from-port="0" to-layer="270" to-port="3"/>
+ <edge from-layer="240" from-port="0" to-layer="270" to-port="4"/>
+ <edge from-layer="271" from-port="0" to-layer="272" to-port="0"/>
+ <edge from-layer="272" from-port="1" to-layer="274" to-port="0"/>
+ <edge from-layer="273" from-port="0" to-layer="274" to-port="1"/>
+ <edge from-layer="274" from-port="2" to-layer="276" to-port="0"/>
+ <edge from-layer="275" from-port="0" to-layer="276" to-port="1"/>
+ <edge from-layer="270" from-port="5" to-layer="277" to-port="0"/>
+ <edge from-layer="276" from-port="2" to-layer="277" to-port="1"/>
+ <edge from-layer="277" from-port="2" to-layer="279" to-port="0"/>
+ <edge from-layer="278" from-port="0" to-layer="279" to-port="1"/>
+ <edge from-layer="279" from-port="2" to-layer="280" to-port="0"/>
+ <edge from-layer="233" from-port="0" to-layer="280" to-port="1"/>
+ <edge from-layer="234" from-port="0" to-layer="280" to-port="2"/>
+ <edge from-layer="235" from-port="0" to-layer="280" to-port="3"/>
+ <edge from-layer="236" from-port="0" to-layer="280" to-port="4"/>
+ <edge from-layer="232" from-port="5" to-layer="281" to-port="0"/>
+ <edge from-layer="280" from-port="5" to-layer="281" to-port="1"/>
+ <edge from-layer="281" from-port="2" to-layer="283" to-port="0"/>
+ <edge from-layer="282" from-port="0" to-layer="283" to-port="1"/>
+ <edge from-layer="283" from-port="2" to-layer="284" to-port="0"/>
+ <edge from-layer="148" from-port="0" to-layer="284" to-port="1"/>
+ <edge from-layer="149" from-port="0" to-layer="284" to-port="2"/>
+ <edge from-layer="150" from-port="0" to-layer="284" to-port="3"/>
+ <edge from-layer="151" from-port="0" to-layer="284" to-port="4"/>
+ <edge from-layer="297" from-port="0" to-layer="298" to-port="0"/>
+ <edge from-layer="298" from-port="1" to-layer="300" to-port="0"/>
+ <edge from-layer="299" from-port="0" to-layer="300" to-port="1"/>
+ <edge from-layer="300" from-port="2" to-layer="302" to-port="0"/>
+ <edge from-layer="301" from-port="0" to-layer="302" to-port="1"/>
+ <edge from-layer="284" from-port="5" to-layer="303" to-port="0"/>
+ <edge from-layer="302" from-port="2" to-layer="303" to-port="1"/>
+ <edge from-layer="303" from-port="2" to-layer="305" to-port="0"/>
+ <edge from-layer="304" from-port="0" to-layer="305" to-port="1"/>
+ <edge from-layer="305" from-port="2" to-layer="307" to-port="0"/>
+ <edge from-layer="306" from-port="0" to-layer="307" to-port="1"/>
+ <edge from-layer="307" from-port="2" to-layer="308" to-port="0"/>
+ <edge from-layer="293" from-port="0" to-layer="308" to-port="1"/>
+ <edge from-layer="294" from-port="0" to-layer="308" to-port="2"/>
+ <edge from-layer="295" from-port="0" to-layer="308" to-port="3"/>
+ <edge from-layer="296" from-port="0" to-layer="308" to-port="4"/>
+ <edge from-layer="310" from-port="0" to-layer="311" to-port="0"/>
+ <edge from-layer="311" from-port="1" to-layer="313" to-port="0"/>
+ <edge from-layer="312" from-port="0" to-layer="313" to-port="1"/>
+ <edge from-layer="313" from-port="2" to-layer="315" to-port="0"/>
+ <edge from-layer="314" from-port="0" to-layer="315" to-port="1"/>
+ <edge from-layer="315" from-port="2" to-layer="316" to-port="0"/>
+ <edge from-layer="309" from-port="0" to-layer="316" to-port="1"/>
+ <edge from-layer="308" from-port="5" to-layer="317" to-port="0"/>
+ <edge from-layer="316" from-port="2" to-layer="317" to-port="1"/>
+ <edge from-layer="317" from-port="2" to-layer="319" to-port="0"/>
+ <edge from-layer="318" from-port="0" to-layer="319" to-port="1"/>
+ <edge from-layer="319" from-port="2" to-layer="321" to-port="0"/>
+ <edge from-layer="320" from-port="0" to-layer="321" to-port="1"/>
+ <edge from-layer="321" from-port="2" to-layer="322" to-port="0"/>
+ <edge from-layer="289" from-port="0" to-layer="322" to-port="1"/>
+ <edge from-layer="290" from-port="0" to-layer="322" to-port="2"/>
+ <edge from-layer="291" from-port="0" to-layer="322" to-port="3"/>
+ <edge from-layer="292" from-port="0" to-layer="322" to-port="4"/>
+ <edge from-layer="323" from-port="0" to-layer="324" to-port="0"/>
+ <edge from-layer="324" from-port="1" to-layer="326" to-port="0"/>
+ <edge from-layer="325" from-port="0" to-layer="326" to-port="1"/>
+ <edge from-layer="326" from-port="2" to-layer="328" to-port="0"/>
+ <edge from-layer="327" from-port="0" to-layer="328" to-port="1"/>
+ <edge from-layer="322" from-port="5" to-layer="329" to-port="0"/>
+ <edge from-layer="328" from-port="2" to-layer="329" to-port="1"/>
+ <edge from-layer="329" from-port="2" to-layer="331" to-port="0"/>
+ <edge from-layer="330" from-port="0" to-layer="331" to-port="1"/>
+ <edge from-layer="331" from-port="2" to-layer="332" to-port="0"/>
+ <edge from-layer="285" from-port="0" to-layer="332" to-port="1"/>
+ <edge from-layer="286" from-port="0" to-layer="332" to-port="2"/>
+ <edge from-layer="287" from-port="0" to-layer="332" to-port="3"/>
+ <edge from-layer="288" from-port="0" to-layer="332" to-port="4"/>
+ <edge from-layer="284" from-port="5" to-layer="333" to-port="0"/>
+ <edge from-layer="332" from-port="5" to-layer="333" to-port="1"/>
+ <edge from-layer="333" from-port="2" to-layer="335" to-port="0"/>
+ <edge from-layer="334" from-port="0" to-layer="335" to-port="1"/>
+ <edge from-layer="335" from-port="2" to-layer="336" to-port="0"/>
+ <edge from-layer="144" from-port="0" to-layer="336" to-port="1"/>
+ <edge from-layer="145" from-port="0" to-layer="336" to-port="2"/>
+ <edge from-layer="146" from-port="0" to-layer="336" to-port="3"/>
+ <edge from-layer="147" from-port="0" to-layer="336" to-port="4"/>
+ <edge from-layer="349" from-port="0" to-layer="350" to-port="0"/>
+ <edge from-layer="350" from-port="1" to-layer="352" to-port="0"/>
+ <edge from-layer="351" from-port="0" to-layer="352" to-port="1"/>
+ <edge from-layer="352" from-port="2" to-layer="354" to-port="0"/>
+ <edge from-layer="353" from-port="0" to-layer="354" to-port="1"/>
+ <edge from-layer="336" from-port="5" to-layer="355" to-port="0"/>
+ <edge from-layer="354" from-port="2" to-layer="355" to-port="1"/>
+ <edge from-layer="355" from-port="2" to-layer="357" to-port="0"/>
+ <edge from-layer="356" from-port="0" to-layer="357" to-port="1"/>
+ <edge from-layer="357" from-port="2" to-layer="359" to-port="0"/>
+ <edge from-layer="358" from-port="0" to-layer="359" to-port="1"/>
+ <edge from-layer="359" from-port="2" to-layer="360" to-port="0"/>
+ <edge from-layer="345" from-port="0" to-layer="360" to-port="1"/>
+ <edge from-layer="346" from-port="0" to-layer="360" to-port="2"/>
+ <edge from-layer="347" from-port="0" to-layer="360" to-port="3"/>
+ <edge from-layer="348" from-port="0" to-layer="360" to-port="4"/>
+ <edge from-layer="362" from-port="0" to-layer="363" to-port="0"/>
+ <edge from-layer="363" from-port="1" to-layer="365" to-port="0"/>
+ <edge from-layer="364" from-port="0" to-layer="365" to-port="1"/>
+ <edge from-layer="365" from-port="2" to-layer="367" to-port="0"/>
+ <edge from-layer="366" from-port="0" to-layer="367" to-port="1"/>
+ <edge from-layer="367" from-port="2" to-layer="368" to-port="0"/>
+ <edge from-layer="361" from-port="0" to-layer="368" to-port="1"/>
+ <edge from-layer="360" from-port="5" to-layer="369" to-port="0"/>
+ <edge from-layer="368" from-port="2" to-layer="369" to-port="1"/>
+ <edge from-layer="369" from-port="2" to-layer="371" to-port="0"/>
+ <edge from-layer="370" from-port="0" to-layer="371" to-port="1"/>
+ <edge from-layer="371" from-port="2" to-layer="373" to-port="0"/>
+ <edge from-layer="372" from-port="0" to-layer="373" to-port="1"/>
+ <edge from-layer="373" from-port="2" to-layer="374" to-port="0"/>
+ <edge from-layer="341" from-port="0" to-layer="374" to-port="1"/>
+ <edge from-layer="342" from-port="0" to-layer="374" to-port="2"/>
+ <edge from-layer="343" from-port="0" to-layer="374" to-port="3"/>
+ <edge from-layer="344" from-port="0" to-layer="374" to-port="4"/>
+ <edge from-layer="375" from-port="0" to-layer="376" to-port="0"/>
+ <edge from-layer="376" from-port="1" to-layer="378" to-port="0"/>
+ <edge from-layer="377" from-port="0" to-layer="378" to-port="1"/>
+ <edge from-layer="378" from-port="2" to-layer="380" to-port="0"/>
+ <edge from-layer="379" from-port="0" to-layer="380" to-port="1"/>
+ <edge from-layer="374" from-port="5" to-layer="381" to-port="0"/>
+ <edge from-layer="380" from-port="2" to-layer="381" to-port="1"/>
+ <edge from-layer="381" from-port="2" to-layer="383" to-port="0"/>
+ <edge from-layer="382" from-port="0" to-layer="383" to-port="1"/>
+ <edge from-layer="383" from-port="2" to-layer="384" to-port="0"/>
+ <edge from-layer="337" from-port="0" to-layer="384" to-port="1"/>
+ <edge from-layer="338" from-port="0" to-layer="384" to-port="2"/>
+ <edge from-layer="339" from-port="0" to-layer="384" to-port="3"/>
+ <edge from-layer="340" from-port="0" to-layer="384" to-port="4"/>
+ <edge from-layer="336" from-port="5" to-layer="385" to-port="0"/>
+ <edge from-layer="384" from-port="5" to-layer="385" to-port="1"/>
+ <edge from-layer="385" from-port="2" to-layer="387" to-port="0"/>
+ <edge from-layer="386" from-port="0" to-layer="387" to-port="1"/>
+ <edge from-layer="387" from-port="2" to-layer="388" to-port="0"/>
+ <edge from-layer="140" from-port="0" to-layer="388" to-port="1"/>
+ <edge from-layer="141" from-port="0" to-layer="388" to-port="2"/>
+ <edge from-layer="142" from-port="0" to-layer="388" to-port="3"/>
+ <edge from-layer="143" from-port="0" to-layer="388" to-port="4"/>
+ <edge from-layer="388" from-port="5" to-layer="389" to-port="0"/>
+ <edge from-layer="390" from-port="0" to-layer="391" to-port="0"/>
+ <edge from-layer="391" from-port="1" to-layer="393" to-port="0"/>
+ <edge from-layer="392" from-port="0" to-layer="393" to-port="1"/>
+ <edge from-layer="393" from-port="2" to-layer="395" to-port="0"/>
+ <edge from-layer="394" from-port="0" to-layer="395" to-port="1"/>
+ <edge from-layer="389" from-port="1" to-layer="396" to-port="0"/>
+ <edge from-layer="395" from-port="2" to-layer="396" to-port="1"/>
+ <edge from-layer="396" from-port="2" to-layer="398" to-port="0"/>
+ <edge from-layer="397" from-port="0" to-layer="398" to-port="1"/>
+ <edge from-layer="398" from-port="2" to-layer="399" to-port="0"/>
+ <edge from-layer="136" from-port="0" to-layer="399" to-port="1"/>
+ <edge from-layer="137" from-port="0" to-layer="399" to-port="2"/>
+ <edge from-layer="138" from-port="0" to-layer="399" to-port="3"/>
+ <edge from-layer="139" from-port="0" to-layer="399" to-port="4"/>
+ <edge from-layer="412" from-port="0" to-layer="413" to-port="0"/>
+ <edge from-layer="413" from-port="1" to-layer="415" to-port="0"/>
+ <edge from-layer="414" from-port="0" to-layer="415" to-port="1"/>
+ <edge from-layer="415" from-port="2" to-layer="417" to-port="0"/>
+ <edge from-layer="416" from-port="0" to-layer="417" to-port="1"/>
+ <edge from-layer="388" from-port="5" to-layer="418" to-port="0"/>
+ <edge from-layer="417" from-port="2" to-layer="418" to-port="1"/>
+ <edge from-layer="418" from-port="2" to-layer="420" to-port="0"/>
+ <edge from-layer="419" from-port="0" to-layer="420" to-port="1"/>
+ <edge from-layer="420" from-port="2" to-layer="422" to-port="0"/>
+ <edge from-layer="421" from-port="0" to-layer="422" to-port="1"/>
+ <edge from-layer="422" from-port="2" to-layer="423" to-port="0"/>
+ <edge from-layer="408" from-port="0" to-layer="423" to-port="1"/>
+ <edge from-layer="409" from-port="0" to-layer="423" to-port="2"/>
+ <edge from-layer="410" from-port="0" to-layer="423" to-port="3"/>
+ <edge from-layer="411" from-port="0" to-layer="423" to-port="4"/>
+ <edge from-layer="425" from-port="0" to-layer="426" to-port="0"/>
+ <edge from-layer="426" from-port="1" to-layer="428" to-port="0"/>
+ <edge from-layer="427" from-port="0" to-layer="428" to-port="1"/>
+ <edge from-layer="428" from-port="2" to-layer="430" to-port="0"/>
+ <edge from-layer="429" from-port="0" to-layer="430" to-port="1"/>
+ <edge from-layer="430" from-port="2" to-layer="431" to-port="0"/>
+ <edge from-layer="424" from-port="0" to-layer="431" to-port="1"/>
+ <edge from-layer="423" from-port="5" to-layer="432" to-port="0"/>
+ <edge from-layer="431" from-port="2" to-layer="432" to-port="1"/>
+ <edge from-layer="432" from-port="2" to-layer="434" to-port="0"/>
+ <edge from-layer="433" from-port="0" to-layer="434" to-port="1"/>
+ <edge from-layer="434" from-port="2" to-layer="436" to-port="0"/>
+ <edge from-layer="435" from-port="0" to-layer="436" to-port="1"/>
+ <edge from-layer="436" from-port="2" to-layer="437" to-port="0"/>
+ <edge from-layer="404" from-port="0" to-layer="437" to-port="1"/>
+ <edge from-layer="405" from-port="0" to-layer="437" to-port="2"/>
+ <edge from-layer="406" from-port="0" to-layer="437" to-port="3"/>
+ <edge from-layer="407" from-port="0" to-layer="437" to-port="4"/>
+ <edge from-layer="438" from-port="0" to-layer="439" to-port="0"/>
+ <edge from-layer="439" from-port="1" to-layer="441" to-port="0"/>
+ <edge from-layer="440" from-port="0" to-layer="441" to-port="1"/>
+ <edge from-layer="441" from-port="2" to-layer="443" to-port="0"/>
+ <edge from-layer="442" from-port="0" to-layer="443" to-port="1"/>
+ <edge from-layer="437" from-port="5" to-layer="444" to-port="0"/>
+ <edge from-layer="443" from-port="2" to-layer="444" to-port="1"/>
+ <edge from-layer="444" from-port="2" to-layer="446" to-port="0"/>
+ <edge from-layer="445" from-port="0" to-layer="446" to-port="1"/>
+ <edge from-layer="446" from-port="2" to-layer="447" to-port="0"/>
+ <edge from-layer="400" from-port="0" to-layer="447" to-port="1"/>
+ <edge from-layer="401" from-port="0" to-layer="447" to-port="2"/>
+ <edge from-layer="402" from-port="0" to-layer="447" to-port="3"/>
+ <edge from-layer="403" from-port="0" to-layer="447" to-port="4"/>
+ <edge from-layer="399" from-port="5" to-layer="448" to-port="0"/>
+ <edge from-layer="447" from-port="5" to-layer="448" to-port="1"/>
+ <edge from-layer="448" from-port="2" to-layer="450" to-port="0"/>
+ <edge from-layer="449" from-port="0" to-layer="450" to-port="1"/>
+ <edge from-layer="450" from-port="2" to-layer="451" to-port="0"/>
+ <edge from-layer="132" from-port="0" to-layer="451" to-port="1"/>
+ <edge from-layer="133" from-port="0" to-layer="451" to-port="2"/>
+ <edge from-layer="134" from-port="0" to-layer="451" to-port="3"/>
+ <edge from-layer="135" from-port="0" to-layer="451" to-port="4"/>
+ <edge from-layer="464" from-port="0" to-layer="465" to-port="0"/>
+ <edge from-layer="465" from-port="1" to-layer="467" to-port="0"/>
+ <edge from-layer="466" from-port="0" to-layer="467" to-port="1"/>
+ <edge from-layer="467" from-port="2" to-layer="469" to-port="0"/>
+ <edge from-layer="468" from-port="0" to-layer="469" to-port="1"/>
+ <edge from-layer="451" from-port="5" to-layer="470" to-port="0"/>
+ <edge from-layer="469" from-port="2" to-layer="470" to-port="1"/>
+ <edge from-layer="470" from-port="2" to-layer="472" to-port="0"/>
+ <edge from-layer="471" from-port="0" to-layer="472" to-port="1"/>
+ <edge from-layer="472" from-port="2" to-layer="474" to-port="0"/>
+ <edge from-layer="473" from-port="0" to-layer="474" to-port="1"/>
+ <edge from-layer="474" from-port="2" to-layer="475" to-port="0"/>
+ <edge from-layer="460" from-port="0" to-layer="475" to-port="1"/>
+ <edge from-layer="461" from-port="0" to-layer="475" to-port="2"/>
+ <edge from-layer="462" from-port="0" to-layer="475" to-port="3"/>
+ <edge from-layer="463" from-port="0" to-layer="475" to-port="4"/>
+ <edge from-layer="477" from-port="0" to-layer="478" to-port="0"/>
+ <edge from-layer="478" from-port="1" to-layer="480" to-port="0"/>
+ <edge from-layer="479" from-port="0" to-layer="480" to-port="1"/>
+ <edge from-layer="480" from-port="2" to-layer="482" to-port="0"/>
+ <edge from-layer="481" from-port="0" to-layer="482" to-port="1"/>
+ <edge from-layer="482" from-port="2" to-layer="483" to-port="0"/>
+ <edge from-layer="476" from-port="0" to-layer="483" to-port="1"/>
+ <edge from-layer="475" from-port="5" to-layer="484" to-port="0"/>
+ <edge from-layer="483" from-port="2" to-layer="484" to-port="1"/>
+ <edge from-layer="484" from-port="2" to-layer="486" to-port="0"/>
+ <edge from-layer="485" from-port="0" to-layer="486" to-port="1"/>
+ <edge from-layer="486" from-port="2" to-layer="488" to-port="0"/>
+ <edge from-layer="487" from-port="0" to-layer="488" to-port="1"/>
+ <edge from-layer="488" from-port="2" to-layer="489" to-port="0"/>
+ <edge from-layer="456" from-port="0" to-layer="489" to-port="1"/>
+ <edge from-layer="457" from-port="0" to-layer="489" to-port="2"/>
+ <edge from-layer="458" from-port="0" to-layer="489" to-port="3"/>
+ <edge from-layer="459" from-port="0" to-layer="489" to-port="4"/>
+ <edge from-layer="490" from-port="0" to-layer="491" to-port="0"/>
+ <edge from-layer="491" from-port="1" to-layer="493" to-port="0"/>
+ <edge from-layer="492" from-port="0" to-layer="493" to-port="1"/>
+ <edge from-layer="493" from-port="2" to-layer="495" to-port="0"/>
+ <edge from-layer="494" from-port="0" to-layer="495" to-port="1"/>
+ <edge from-layer="489" from-port="5" to-layer="496" to-port="0"/>
+ <edge from-layer="495" from-port="2" to-layer="496" to-port="1"/>
+ <edge from-layer="496" from-port="2" to-layer="498" to-port="0"/>
+ <edge from-layer="497" from-port="0" to-layer="498" to-port="1"/>
+ <edge from-layer="498" from-port="2" to-layer="499" to-port="0"/>
+ <edge from-layer="452" from-port="0" to-layer="499" to-port="1"/>
+ <edge from-layer="453" from-port="0" to-layer="499" to-port="2"/>
+ <edge from-layer="454" from-port="0" to-layer="499" to-port="3"/>
+ <edge from-layer="455" from-port="0" to-layer="499" to-port="4"/>
+ <edge from-layer="451" from-port="5" to-layer="500" to-port="0"/>
+ <edge from-layer="499" from-port="5" to-layer="500" to-port="1"/>
+ <edge from-layer="500" from-port="2" to-layer="502" to-port="0"/>
+ <edge from-layer="501" from-port="0" to-layer="502" to-port="1"/>
+ <edge from-layer="502" from-port="2" to-layer="503" to-port="0"/>
+ <edge from-layer="128" from-port="0" to-layer="503" to-port="1"/>
+ <edge from-layer="129" from-port="0" to-layer="503" to-port="2"/>
+ <edge from-layer="130" from-port="0" to-layer="503" to-port="3"/>
+ <edge from-layer="131" from-port="0" to-layer="503" to-port="4"/>
+ <edge from-layer="516" from-port="0" to-layer="517" to-port="0"/>
+ <edge from-layer="517" from-port="1" to-layer="519" to-port="0"/>
+ <edge from-layer="518" from-port="0" to-layer="519" to-port="1"/>
+ <edge from-layer="519" from-port="2" to-layer="521" to-port="0"/>
+ <edge from-layer="520" from-port="0" to-layer="521" to-port="1"/>
+ <edge from-layer="503" from-port="5" to-layer="522" to-port="0"/>
+ <edge from-layer="521" from-port="2" to-layer="522" to-port="1"/>
+ <edge from-layer="522" from-port="2" to-layer="524" to-port="0"/>
+ <edge from-layer="523" from-port="0" to-layer="524" to-port="1"/>
+ <edge from-layer="524" from-port="2" to-layer="526" to-port="0"/>
+ <edge from-layer="525" from-port="0" to-layer="526" to-port="1"/>
+ <edge from-layer="526" from-port="2" to-layer="527" to-port="0"/>
+ <edge from-layer="512" from-port="0" to-layer="527" to-port="1"/>
+ <edge from-layer="513" from-port="0" to-layer="527" to-port="2"/>
+ <edge from-layer="514" from-port="0" to-layer="527" to-port="3"/>
+ <edge from-layer="515" from-port="0" to-layer="527" to-port="4"/>
+ <edge from-layer="529" from-port="0" to-layer="530" to-port="0"/>
+ <edge from-layer="530" from-port="1" to-layer="532" to-port="0"/>
+ <edge from-layer="531" from-port="0" to-layer="532" to-port="1"/>
+ <edge from-layer="532" from-port="2" to-layer="534" to-port="0"/>
+ <edge from-layer="533" from-port="0" to-layer="534" to-port="1"/>
+ <edge from-layer="534" from-port="2" to-layer="535" to-port="0"/>
+ <edge from-layer="528" from-port="0" to-layer="535" to-port="1"/>
+ <edge from-layer="527" from-port="5" to-layer="536" to-port="0"/>
+ <edge from-layer="535" from-port="2" to-layer="536" to-port="1"/>
+ <edge from-layer="536" from-port="2" to-layer="538" to-port="0"/>
+ <edge from-layer="537" from-port="0" to-layer="538" to-port="1"/>
+ <edge from-layer="538" from-port="2" to-layer="540" to-port="0"/>
+ <edge from-layer="539" from-port="0" to-layer="540" to-port="1"/>
+ <edge from-layer="540" from-port="2" to-layer="541" to-port="0"/>
+ <edge from-layer="508" from-port="0" to-layer="541" to-port="1"/>
+ <edge from-layer="509" from-port="0" to-layer="541" to-port="2"/>
+ <edge from-layer="510" from-port="0" to-layer="541" to-port="3"/>
+ <edge from-layer="511" from-port="0" to-layer="541" to-port="4"/>
+ <edge from-layer="542" from-port="0" to-layer="543" to-port="0"/>
+ <edge from-layer="543" from-port="1" to-layer="545" to-port="0"/>
+ <edge from-layer="544" from-port="0" to-layer="545" to-port="1"/>
+ <edge from-layer="545" from-port="2" to-layer="547" to-port="0"/>
+ <edge from-layer="546" from-port="0" to-layer="547" to-port="1"/>
+ <edge from-layer="541" from-port="5" to-layer="548" to-port="0"/>
+ <edge from-layer="547" from-port="2" to-layer="548" to-port="1"/>
+ <edge from-layer="548" from-port="2" to-layer="550" to-port="0"/>
+ <edge from-layer="549" from-port="0" to-layer="550" to-port="1"/>
+ <edge from-layer="550" from-port="2" to-layer="551" to-port="0"/>
+ <edge from-layer="504" from-port="0" to-layer="551" to-port="1"/>
+ <edge from-layer="505" from-port="0" to-layer="551" to-port="2"/>
+ <edge from-layer="506" from-port="0" to-layer="551" to-port="3"/>
+ <edge from-layer="507" from-port="0" to-layer="551" to-port="4"/>
+ <edge from-layer="503" from-port="5" to-layer="552" to-port="0"/>
+ <edge from-layer="551" from-port="5" to-layer="552" to-port="1"/>
+ <edge from-layer="552" from-port="2" to-layer="554" to-port="0"/>
+ <edge from-layer="553" from-port="0" to-layer="554" to-port="1"/>
+ <edge from-layer="554" from-port="2" to-layer="555" to-port="0"/>
+ <edge from-layer="124" from-port="0" to-layer="555" to-port="1"/>
+ <edge from-layer="125" from-port="0" to-layer="555" to-port="2"/>
+ <edge from-layer="126" from-port="0" to-layer="555" to-port="3"/>
+ <edge from-layer="127" from-port="0" to-layer="555" to-port="4"/>
+ <edge from-layer="568" from-port="0" to-layer="569" to-port="0"/>
+ <edge from-layer="569" from-port="1" to-layer="571" to-port="0"/>
+ <edge from-layer="570" from-port="0" to-layer="571" to-port="1"/>
+ <edge from-layer="571" from-port="2" to-layer="573" to-port="0"/>
+ <edge from-layer="572" from-port="0" to-layer="573" to-port="1"/>
+ <edge from-layer="555" from-port="5" to-layer="574" to-port="0"/>
+ <edge from-layer="573" from-port="2" to-layer="574" to-port="1"/>
+ <edge from-layer="574" from-port="2" to-layer="576" to-port="0"/>
+ <edge from-layer="575" from-port="0" to-layer="576" to-port="1"/>
+ <edge from-layer="576" from-port="2" to-layer="578" to-port="0"/>
+ <edge from-layer="577" from-port="0" to-layer="578" to-port="1"/>
+ <edge from-layer="578" from-port="2" to-layer="579" to-port="0"/>
+ <edge from-layer="564" from-port="0" to-layer="579" to-port="1"/>
+ <edge from-layer="565" from-port="0" to-layer="579" to-port="2"/>
+ <edge from-layer="566" from-port="0" to-layer="579" to-port="3"/>
+ <edge from-layer="567" from-port="0" to-layer="579" to-port="4"/>
+ <edge from-layer="581" from-port="0" to-layer="582" to-port="0"/>
+ <edge from-layer="582" from-port="1" to-layer="584" to-port="0"/>
+ <edge from-layer="583" from-port="0" to-layer="584" to-port="1"/>
+ <edge from-layer="584" from-port="2" to-layer="586" to-port="0"/>
+ <edge from-layer="585" from-port="0" to-layer="586" to-port="1"/>
+ <edge from-layer="586" from-port="2" to-layer="587" to-port="0"/>
+ <edge from-layer="580" from-port="0" to-layer="587" to-port="1"/>
+ <edge from-layer="579" from-port="5" to-layer="588" to-port="0"/>
+ <edge from-layer="587" from-port="2" to-layer="588" to-port="1"/>
+ <edge from-layer="588" from-port="2" to-layer="590" to-port="0"/>
+ <edge from-layer="589" from-port="0" to-layer="590" to-port="1"/>
+ <edge from-layer="590" from-port="2" to-layer="592" to-port="0"/>
+ <edge from-layer="591" from-port="0" to-layer="592" to-port="1"/>
+ <edge from-layer="592" from-port="2" to-layer="593" to-port="0"/>
+ <edge from-layer="560" from-port="0" to-layer="593" to-port="1"/>
+ <edge from-layer="561" from-port="0" to-layer="593" to-port="2"/>
+ <edge from-layer="562" from-port="0" to-layer="593" to-port="3"/>
+ <edge from-layer="563" from-port="0" to-layer="593" to-port="4"/>
+ <edge from-layer="594" from-port="0" to-layer="595" to-port="0"/>
+ <edge from-layer="595" from-port="1" to-layer="597" to-port="0"/>
+ <edge from-layer="596" from-port="0" to-layer="597" to-port="1"/>
+ <edge from-layer="597" from-port="2" to-layer="599" to-port="0"/>
+ <edge from-layer="598" from-port="0" to-layer="599" to-port="1"/>
+ <edge from-layer="593" from-port="5" to-layer="600" to-port="0"/>
+ <edge from-layer="599" from-port="2" to-layer="600" to-port="1"/>
+ <edge from-layer="600" from-port="2" to-layer="602" to-port="0"/>
+ <edge from-layer="601" from-port="0" to-layer="602" to-port="1"/>
+ <edge from-layer="602" from-port="2" to-layer="603" to-port="0"/>
+ <edge from-layer="556" from-port="0" to-layer="603" to-port="1"/>
+ <edge from-layer="557" from-port="0" to-layer="603" to-port="2"/>
+ <edge from-layer="558" from-port="0" to-layer="603" to-port="3"/>
+ <edge from-layer="559" from-port="0" to-layer="603" to-port="4"/>
+ <edge from-layer="555" from-port="5" to-layer="604" to-port="0"/>
+ <edge from-layer="603" from-port="5" to-layer="604" to-port="1"/>
+ <edge from-layer="604" from-port="2" to-layer="606" to-port="0"/>
+ <edge from-layer="605" from-port="0" to-layer="606" to-port="1"/>
+ <edge from-layer="606" from-port="2" to-layer="607" to-port="0"/>
+ <edge from-layer="120" from-port="0" to-layer="607" to-port="1"/>
+ <edge from-layer="121" from-port="0" to-layer="607" to-port="2"/>
+ <edge from-layer="122" from-port="0" to-layer="607" to-port="3"/>
+ <edge from-layer="123" from-port="0" to-layer="607" to-port="4"/>
+ <edge from-layer="620" from-port="0" to-layer="621" to-port="0"/>
+ <edge from-layer="621" from-port="1" to-layer="623" to-port="0"/>
+ <edge from-layer="622" from-port="0" to-layer="623" to-port="1"/>
+ <edge from-layer="623" from-port="2" to-layer="625" to-port="0"/>
+ <edge from-layer="624" from-port="0" to-layer="625" to-port="1"/>
+ <edge from-layer="607" from-port="5" to-layer="626" to-port="0"/>
+ <edge from-layer="625" from-port="2" to-layer="626" to-port="1"/>
+ <edge from-layer="626" from-port="2" to-layer="628" to-port="0"/>
+ <edge from-layer="627" from-port="0" to-layer="628" to-port="1"/>
+ <edge from-layer="628" from-port="2" to-layer="630" to-port="0"/>
+ <edge from-layer="629" from-port="0" to-layer="630" to-port="1"/>
+ <edge from-layer="630" from-port="2" to-layer="631" to-port="0"/>
+ <edge from-layer="616" from-port="0" to-layer="631" to-port="1"/>
+ <edge from-layer="617" from-port="0" to-layer="631" to-port="2"/>
+ <edge from-layer="618" from-port="0" to-layer="631" to-port="3"/>
+ <edge from-layer="619" from-port="0" to-layer="631" to-port="4"/>
+ <edge from-layer="633" from-port="0" to-layer="634" to-port="0"/>
+ <edge from-layer="634" from-port="1" to-layer="636" to-port="0"/>
+ <edge from-layer="635" from-port="0" to-layer="636" to-port="1"/>
+ <edge from-layer="636" from-port="2" to-layer="638" to-port="0"/>
+ <edge from-layer="637" from-port="0" to-layer="638" to-port="1"/>
+ <edge from-layer="638" from-port="2" to-layer="639" to-port="0"/>
+ <edge from-layer="632" from-port="0" to-layer="639" to-port="1"/>
+ <edge from-layer="631" from-port="5" to-layer="640" to-port="0"/>
+ <edge from-layer="639" from-port="2" to-layer="640" to-port="1"/>
+ <edge from-layer="640" from-port="2" to-layer="642" to-port="0"/>
+ <edge from-layer="641" from-port="0" to-layer="642" to-port="1"/>
+ <edge from-layer="642" from-port="2" to-layer="644" to-port="0"/>
+ <edge from-layer="643" from-port="0" to-layer="644" to-port="1"/>
+ <edge from-layer="644" from-port="2" to-layer="645" to-port="0"/>
+ <edge from-layer="612" from-port="0" to-layer="645" to-port="1"/>
+ <edge from-layer="613" from-port="0" to-layer="645" to-port="2"/>
+ <edge from-layer="614" from-port="0" to-layer="645" to-port="3"/>
+ <edge from-layer="615" from-port="0" to-layer="645" to-port="4"/>
+ <edge from-layer="646" from-port="0" to-layer="647" to-port="0"/>
+ <edge from-layer="647" from-port="1" to-layer="649" to-port="0"/>
+ <edge from-layer="648" from-port="0" to-layer="649" to-port="1"/>
+ <edge from-layer="649" from-port="2" to-layer="651" to-port="0"/>
+ <edge from-layer="650" from-port="0" to-layer="651" to-port="1"/>
+ <edge from-layer="645" from-port="5" to-layer="652" to-port="0"/>
+ <edge from-layer="651" from-port="2" to-layer="652" to-port="1"/>
+ <edge from-layer="652" from-port="2" to-layer="654" to-port="0"/>
+ <edge from-layer="653" from-port="0" to-layer="654" to-port="1"/>
+ <edge from-layer="654" from-port="2" to-layer="655" to-port="0"/>
+ <edge from-layer="608" from-port="0" to-layer="655" to-port="1"/>
+ <edge from-layer="609" from-port="0" to-layer="655" to-port="2"/>
+ <edge from-layer="610" from-port="0" to-layer="655" to-port="3"/>
+ <edge from-layer="611" from-port="0" to-layer="655" to-port="4"/>
+ <edge from-layer="607" from-port="5" to-layer="656" to-port="0"/>
+ <edge from-layer="655" from-port="5" to-layer="656" to-port="1"/>
+ <edge from-layer="656" from-port="2" to-layer="658" to-port="0"/>
+ <edge from-layer="657" from-port="0" to-layer="658" to-port="1"/>
+ <edge from-layer="658" from-port="2" to-layer="659" to-port="0"/>
+ <edge from-layer="116" from-port="0" to-layer="659" to-port="1"/>
+ <edge from-layer="117" from-port="0" to-layer="659" to-port="2"/>
+ <edge from-layer="118" from-port="0" to-layer="659" to-port="3"/>
+ <edge from-layer="119" from-port="0" to-layer="659" to-port="4"/>
+ <edge from-layer="672" from-port="0" to-layer="673" to-port="0"/>
+ <edge from-layer="673" from-port="1" to-layer="675" to-port="0"/>
+ <edge from-layer="674" from-port="0" to-layer="675" to-port="1"/>
+ <edge from-layer="675" from-port="2" to-layer="677" to-port="0"/>
+ <edge from-layer="676" from-port="0" to-layer="677" to-port="1"/>
+ <edge from-layer="659" from-port="5" to-layer="678" to-port="0"/>
+ <edge from-layer="677" from-port="2" to-layer="678" to-port="1"/>
+ <edge from-layer="678" from-port="2" to-layer="680" to-port="0"/>
+ <edge from-layer="679" from-port="0" to-layer="680" to-port="1"/>
+ <edge from-layer="680" from-port="2" to-layer="682" to-port="0"/>
+ <edge from-layer="681" from-port="0" to-layer="682" to-port="1"/>
+ <edge from-layer="682" from-port="2" to-layer="683" to-port="0"/>
+ <edge from-layer="668" from-port="0" to-layer="683" to-port="1"/>
+ <edge from-layer="669" from-port="0" to-layer="683" to-port="2"/>
+ <edge from-layer="670" from-port="0" to-layer="683" to-port="3"/>
+ <edge from-layer="671" from-port="0" to-layer="683" to-port="4"/>
+ <edge from-layer="685" from-port="0" to-layer="686" to-port="0"/>
+ <edge from-layer="686" from-port="1" to-layer="688" to-port="0"/>
+ <edge from-layer="687" from-port="0" to-layer="688" to-port="1"/>
+ <edge from-layer="688" from-port="2" to-layer="690" to-port="0"/>
+ <edge from-layer="689" from-port="0" to-layer="690" to-port="1"/>
+ <edge from-layer="690" from-port="2" to-layer="691" to-port="0"/>
+ <edge from-layer="684" from-port="0" to-layer="691" to-port="1"/>
+ <edge from-layer="683" from-port="5" to-layer="692" to-port="0"/>
+ <edge from-layer="691" from-port="2" to-layer="692" to-port="1"/>
+ <edge from-layer="692" from-port="2" to-layer="694" to-port="0"/>
+ <edge from-layer="693" from-port="0" to-layer="694" to-port="1"/>
+ <edge from-layer="694" from-port="2" to-layer="696" to-port="0"/>
+ <edge from-layer="695" from-port="0" to-layer="696" to-port="1"/>
+ <edge from-layer="696" from-port="2" to-layer="697" to-port="0"/>
+ <edge from-layer="664" from-port="0" to-layer="697" to-port="1"/>
+ <edge from-layer="665" from-port="0" to-layer="697" to-port="2"/>
+ <edge from-layer="666" from-port="0" to-layer="697" to-port="3"/>
+ <edge from-layer="667" from-port="0" to-layer="697" to-port="4"/>
+ <edge from-layer="698" from-port="0" to-layer="699" to-port="0"/>
+ <edge from-layer="699" from-port="1" to-layer="701" to-port="0"/>
+ <edge from-layer="700" from-port="0" to-layer="701" to-port="1"/>
+ <edge from-layer="701" from-port="2" to-layer="703" to-port="0"/>
+ <edge from-layer="702" from-port="0" to-layer="703" to-port="1"/>
+ <edge from-layer="697" from-port="5" to-layer="704" to-port="0"/>
+ <edge from-layer="703" from-port="2" to-layer="704" to-port="1"/>
+ <edge from-layer="704" from-port="2" to-layer="706" to-port="0"/>
+ <edge from-layer="705" from-port="0" to-layer="706" to-port="1"/>
+ <edge from-layer="706" from-port="2" to-layer="707" to-port="0"/>
+ <edge from-layer="660" from-port="0" to-layer="707" to-port="1"/>
+ <edge from-layer="661" from-port="0" to-layer="707" to-port="2"/>
+ <edge from-layer="662" from-port="0" to-layer="707" to-port="3"/>
+ <edge from-layer="663" from-port="0" to-layer="707" to-port="4"/>
+ <edge from-layer="659" from-port="5" to-layer="708" to-port="0"/>
+ <edge from-layer="707" from-port="5" to-layer="708" to-port="1"/>
+ <edge from-layer="708" from-port="2" to-layer="710" to-port="0"/>
+ <edge from-layer="709" from-port="0" to-layer="710" to-port="1"/>
+ <edge from-layer="710" from-port="2" to-layer="711" to-port="0"/>
+ <edge from-layer="112" from-port="0" to-layer="711" to-port="1"/>
+ <edge from-layer="113" from-port="0" to-layer="711" to-port="2"/>
+ <edge from-layer="114" from-port="0" to-layer="711" to-port="3"/>
+ <edge from-layer="115" from-port="0" to-layer="711" to-port="4"/>
+ <edge from-layer="724" from-port="0" to-layer="725" to-port="0"/>
+ <edge from-layer="725" from-port="1" to-layer="727" to-port="0"/>
+ <edge from-layer="726" from-port="0" to-layer="727" to-port="1"/>
+ <edge from-layer="727" from-port="2" to-layer="729" to-port="0"/>
+ <edge from-layer="728" from-port="0" to-layer="729" to-port="1"/>
+ <edge from-layer="711" from-port="5" to-layer="730" to-port="0"/>
+ <edge from-layer="729" from-port="2" to-layer="730" to-port="1"/>
+ <edge from-layer="730" from-port="2" to-layer="732" to-port="0"/>
+ <edge from-layer="731" from-port="0" to-layer="732" to-port="1"/>
+ <edge from-layer="732" from-port="2" to-layer="734" to-port="0"/>
+ <edge from-layer="733" from-port="0" to-layer="734" to-port="1"/>
+ <edge from-layer="734" from-port="2" to-layer="735" to-port="0"/>
+ <edge from-layer="720" from-port="0" to-layer="735" to-port="1"/>
+ <edge from-layer="721" from-port="0" to-layer="735" to-port="2"/>
+ <edge from-layer="722" from-port="0" to-layer="735" to-port="3"/>
+ <edge from-layer="723" from-port="0" to-layer="735" to-port="4"/>
+ <edge from-layer="737" from-port="0" to-layer="738" to-port="0"/>
+ <edge from-layer="738" from-port="1" to-layer="740" to-port="0"/>
+ <edge from-layer="739" from-port="0" to-layer="740" to-port="1"/>
+ <edge from-layer="740" from-port="2" to-layer="742" to-port="0"/>
+ <edge from-layer="741" from-port="0" to-layer="742" to-port="1"/>
+ <edge from-layer="742" from-port="2" to-layer="743" to-port="0"/>
+ <edge from-layer="736" from-port="0" to-layer="743" to-port="1"/>
+ <edge from-layer="735" from-port="5" to-layer="744" to-port="0"/>
+ <edge from-layer="743" from-port="2" to-layer="744" to-port="1"/>
+ <edge from-layer="744" from-port="2" to-layer="746" to-port="0"/>
+ <edge from-layer="745" from-port="0" to-layer="746" to-port="1"/>
+ <edge from-layer="746" from-port="2" to-layer="748" to-port="0"/>
+ <edge from-layer="747" from-port="0" to-layer="748" to-port="1"/>
+ <edge from-layer="748" from-port="2" to-layer="749" to-port="0"/>
+ <edge from-layer="716" from-port="0" to-layer="749" to-port="1"/>
+ <edge from-layer="717" from-port="0" to-layer="749" to-port="2"/>
+ <edge from-layer="718" from-port="0" to-layer="749" to-port="3"/>
+ <edge from-layer="719" from-port="0" to-layer="749" to-port="4"/>
+ <edge from-layer="750" from-port="0" to-layer="751" to-port="0"/>
+ <edge from-layer="751" from-port="1" to-layer="753" to-port="0"/>
+ <edge from-layer="752" from-port="0" to-layer="753" to-port="1"/>
+ <edge from-layer="753" from-port="2" to-layer="755" to-port="0"/>
+ <edge from-layer="754" from-port="0" to-layer="755" to-port="1"/>
+ <edge from-layer="749" from-port="5" to-layer="756" to-port="0"/>
+ <edge from-layer="755" from-port="2" to-layer="756" to-port="1"/>
+ <edge from-layer="756" from-port="2" to-layer="758" to-port="0"/>
+ <edge from-layer="757" from-port="0" to-layer="758" to-port="1"/>
+ <edge from-layer="758" from-port="2" to-layer="759" to-port="0"/>
+ <edge from-layer="712" from-port="0" to-layer="759" to-port="1"/>
+ <edge from-layer="713" from-port="0" to-layer="759" to-port="2"/>
+ <edge from-layer="714" from-port="0" to-layer="759" to-port="3"/>
+ <edge from-layer="715" from-port="0" to-layer="759" to-port="4"/>
+ <edge from-layer="711" from-port="5" to-layer="760" to-port="0"/>
+ <edge from-layer="759" from-port="5" to-layer="760" to-port="1"/>
+ <edge from-layer="760" from-port="2" to-layer="762" to-port="0"/>
+ <edge from-layer="761" from-port="0" to-layer="762" to-port="1"/>
+ <edge from-layer="762" from-port="2" to-layer="763" to-port="0"/>
+ <edge from-layer="108" from-port="0" to-layer="763" to-port="1"/>
+ <edge from-layer="109" from-port="0" to-layer="763" to-port="2"/>
+ <edge from-layer="110" from-port="0" to-layer="763" to-port="3"/>
+ <edge from-layer="111" from-port="0" to-layer="763" to-port="4"/>
+ <edge from-layer="776" from-port="0" to-layer="777" to-port="0"/>
+ <edge from-layer="777" from-port="1" to-layer="779" to-port="0"/>
+ <edge from-layer="778" from-port="0" to-layer="779" to-port="1"/>
+ <edge from-layer="779" from-port="2" to-layer="781" to-port="0"/>
+ <edge from-layer="780" from-port="0" to-layer="781" to-port="1"/>
+ <edge from-layer="763" from-port="5" to-layer="782" to-port="0"/>
+ <edge from-layer="781" from-port="2" to-layer="782" to-port="1"/>
+ <edge from-layer="782" from-port="2" to-layer="784" to-port="0"/>
+ <edge from-layer="783" from-port="0" to-layer="784" to-port="1"/>
+ <edge from-layer="784" from-port="2" to-layer="786" to-port="0"/>
+ <edge from-layer="785" from-port="0" to-layer="786" to-port="1"/>
+ <edge from-layer="786" from-port="2" to-layer="787" to-port="0"/>
+ <edge from-layer="772" from-port="0" to-layer="787" to-port="1"/>
+ <edge from-layer="773" from-port="0" to-layer="787" to-port="2"/>
+ <edge from-layer="774" from-port="0" to-layer="787" to-port="3"/>
+ <edge from-layer="775" from-port="0" to-layer="787" to-port="4"/>
+ <edge from-layer="789" from-port="0" to-layer="790" to-port="0"/>
+ <edge from-layer="790" from-port="1" to-layer="792" to-port="0"/>
+ <edge from-layer="791" from-port="0" to-layer="792" to-port="1"/>
+ <edge from-layer="792" from-port="2" to-layer="794" to-port="0"/>
+ <edge from-layer="793" from-port="0" to-layer="794" to-port="1"/>
+ <edge from-layer="794" from-port="2" to-layer="795" to-port="0"/>
+ <edge from-layer="788" from-port="0" to-layer="795" to-port="1"/>
+ <edge from-layer="787" from-port="5" to-layer="796" to-port="0"/>
+ <edge from-layer="795" from-port="2" to-layer="796" to-port="1"/>
+ <edge from-layer="796" from-port="2" to-layer="798" to-port="0"/>
+ <edge from-layer="797" from-port="0" to-layer="798" to-port="1"/>
+ <edge from-layer="798" from-port="2" to-layer="800" to-port="0"/>
+ <edge from-layer="799" from-port="0" to-layer="800" to-port="1"/>
+ <edge from-layer="800" from-port="2" to-layer="801" to-port="0"/>
+ <edge from-layer="768" from-port="0" to-layer="801" to-port="1"/>
+ <edge from-layer="769" from-port="0" to-layer="801" to-port="2"/>
+ <edge from-layer="770" from-port="0" to-layer="801" to-port="3"/>
+ <edge from-layer="771" from-port="0" to-layer="801" to-port="4"/>
+ <edge from-layer="802" from-port="0" to-layer="803" to-port="0"/>
+ <edge from-layer="803" from-port="1" to-layer="805" to-port="0"/>
+ <edge from-layer="804" from-port="0" to-layer="805" to-port="1"/>
+ <edge from-layer="805" from-port="2" to-layer="807" to-port="0"/>
+ <edge from-layer="806" from-port="0" to-layer="807" to-port="1"/>
+ <edge from-layer="801" from-port="5" to-layer="808" to-port="0"/>
+ <edge from-layer="807" from-port="2" to-layer="808" to-port="1"/>
+ <edge from-layer="808" from-port="2" to-layer="810" to-port="0"/>
+ <edge from-layer="809" from-port="0" to-layer="810" to-port="1"/>
+ <edge from-layer="810" from-port="2" to-layer="811" to-port="0"/>
+ <edge from-layer="764" from-port="0" to-layer="811" to-port="1"/>
+ <edge from-layer="765" from-port="0" to-layer="811" to-port="2"/>
+ <edge from-layer="766" from-port="0" to-layer="811" to-port="3"/>
+ <edge from-layer="767" from-port="0" to-layer="811" to-port="4"/>
+ <edge from-layer="763" from-port="5" to-layer="812" to-port="0"/>
+ <edge from-layer="811" from-port="5" to-layer="812" to-port="1"/>
+ <edge from-layer="812" from-port="2" to-layer="814" to-port="0"/>
+ <edge from-layer="813" from-port="0" to-layer="814" to-port="1"/>
+ <edge from-layer="814" from-port="2" to-layer="815" to-port="0"/>
+ <edge from-layer="104" from-port="0" to-layer="815" to-port="1"/>
+ <edge from-layer="105" from-port="0" to-layer="815" to-port="2"/>
+ <edge from-layer="106" from-port="0" to-layer="815" to-port="3"/>
+ <edge from-layer="107" from-port="0" to-layer="815" to-port="4"/>
+ <edge from-layer="828" from-port="0" to-layer="829" to-port="0"/>
+ <edge from-layer="829" from-port="1" to-layer="831" to-port="0"/>
+ <edge from-layer="830" from-port="0" to-layer="831" to-port="1"/>
+ <edge from-layer="831" from-port="2" to-layer="833" to-port="0"/>
+ <edge from-layer="832" from-port="0" to-layer="833" to-port="1"/>
+ <edge from-layer="815" from-port="5" to-layer="834" to-port="0"/>
+ <edge from-layer="833" from-port="2" to-layer="834" to-port="1"/>
+ <edge from-layer="834" from-port="2" to-layer="836" to-port="0"/>
+ <edge from-layer="835" from-port="0" to-layer="836" to-port="1"/>
+ <edge from-layer="836" from-port="2" to-layer="838" to-port="0"/>
+ <edge from-layer="837" from-port="0" to-layer="838" to-port="1"/>
+ <edge from-layer="838" from-port="2" to-layer="839" to-port="0"/>
+ <edge from-layer="824" from-port="0" to-layer="839" to-port="1"/>
+ <edge from-layer="825" from-port="0" to-layer="839" to-port="2"/>
+ <edge from-layer="826" from-port="0" to-layer="839" to-port="3"/>
+ <edge from-layer="827" from-port="0" to-layer="839" to-port="4"/>
+ <edge from-layer="841" from-port="0" to-layer="842" to-port="0"/>
+ <edge from-layer="842" from-port="1" to-layer="844" to-port="0"/>
+ <edge from-layer="843" from-port="0" to-layer="844" to-port="1"/>
+ <edge from-layer="844" from-port="2" to-layer="846" to-port="0"/>
+ <edge from-layer="845" from-port="0" to-layer="846" to-port="1"/>
+ <edge from-layer="846" from-port="2" to-layer="847" to-port="0"/>
+ <edge from-layer="840" from-port="0" to-layer="847" to-port="1"/>
+ <edge from-layer="839" from-port="5" to-layer="848" to-port="0"/>
+ <edge from-layer="847" from-port="2" to-layer="848" to-port="1"/>
+ <edge from-layer="848" from-port="2" to-layer="850" to-port="0"/>
+ <edge from-layer="849" from-port="0" to-layer="850" to-port="1"/>
+ <edge from-layer="850" from-port="2" to-layer="852" to-port="0"/>
+ <edge from-layer="851" from-port="0" to-layer="852" to-port="1"/>
+ <edge from-layer="852" from-port="2" to-layer="853" to-port="0"/>
+ <edge from-layer="820" from-port="0" to-layer="853" to-port="1"/>
+ <edge from-layer="821" from-port="0" to-layer="853" to-port="2"/>
+ <edge from-layer="822" from-port="0" to-layer="853" to-port="3"/>
+ <edge from-layer="823" from-port="0" to-layer="853" to-port="4"/>
+ <edge from-layer="854" from-port="0" to-layer="855" to-port="0"/>
+ <edge from-layer="855" from-port="1" to-layer="857" to-port="0"/>
+ <edge from-layer="856" from-port="0" to-layer="857" to-port="1"/>
+ <edge from-layer="857" from-port="2" to-layer="859" to-port="0"/>
+ <edge from-layer="858" from-port="0" to-layer="859" to-port="1"/>
+ <edge from-layer="853" from-port="5" to-layer="860" to-port="0"/>
+ <edge from-layer="859" from-port="2" to-layer="860" to-port="1"/>
+ <edge from-layer="860" from-port="2" to-layer="862" to-port="0"/>
+ <edge from-layer="861" from-port="0" to-layer="862" to-port="1"/>
+ <edge from-layer="862" from-port="2" to-layer="863" to-port="0"/>
+ <edge from-layer="816" from-port="0" to-layer="863" to-port="1"/>
+ <edge from-layer="817" from-port="0" to-layer="863" to-port="2"/>
+ <edge from-layer="818" from-port="0" to-layer="863" to-port="3"/>
+ <edge from-layer="819" from-port="0" to-layer="863" to-port="4"/>
+ <edge from-layer="815" from-port="5" to-layer="864" to-port="0"/>
+ <edge from-layer="863" from-port="5" to-layer="864" to-port="1"/>
+ <edge from-layer="864" from-port="2" to-layer="866" to-port="0"/>
+ <edge from-layer="865" from-port="0" to-layer="866" to-port="1"/>
+ <edge from-layer="866" from-port="2" to-layer="867" to-port="0"/>
+ <edge from-layer="100" from-port="0" to-layer="867" to-port="1"/>
+ <edge from-layer="101" from-port="0" to-layer="867" to-port="2"/>
+ <edge from-layer="102" from-port="0" to-layer="867" to-port="3"/>
+ <edge from-layer="103" from-port="0" to-layer="867" to-port="4"/>
+ <edge from-layer="867" from-port="5" to-layer="868" to-port="0"/>
+ <edge from-layer="869" from-port="0" to-layer="870" to-port="0"/>
+ <edge from-layer="870" from-port="1" to-layer="872" to-port="0"/>
+ <edge from-layer="871" from-port="0" to-layer="872" to-port="1"/>
+ <edge from-layer="872" from-port="2" to-layer="874" to-port="0"/>
+ <edge from-layer="873" from-port="0" to-layer="874" to-port="1"/>
+ <edge from-layer="868" from-port="1" to-layer="875" to-port="0"/>
+ <edge from-layer="874" from-port="2" to-layer="875" to-port="1"/>
+ <edge from-layer="875" from-port="2" to-layer="877" to-port="0"/>
+ <edge from-layer="876" from-port="0" to-layer="877" to-port="1"/>
+ <edge from-layer="877" from-port="2" to-layer="878" to-port="0"/>
+ <edge from-layer="96" from-port="0" to-layer="878" to-port="1"/>
+ <edge from-layer="97" from-port="0" to-layer="878" to-port="2"/>
+ <edge from-layer="98" from-port="0" to-layer="878" to-port="3"/>
+ <edge from-layer="99" from-port="0" to-layer="878" to-port="4"/>
+ <edge from-layer="891" from-port="0" to-layer="892" to-port="0"/>
+ <edge from-layer="892" from-port="1" to-layer="894" to-port="0"/>
+ <edge from-layer="893" from-port="0" to-layer="894" to-port="1"/>
+ <edge from-layer="894" from-port="2" to-layer="896" to-port="0"/>
+ <edge from-layer="895" from-port="0" to-layer="896" to-port="1"/>
+ <edge from-layer="867" from-port="5" to-layer="897" to-port="0"/>
+ <edge from-layer="896" from-port="2" to-layer="897" to-port="1"/>
+ <edge from-layer="897" from-port="2" to-layer="899" to-port="0"/>
+ <edge from-layer="898" from-port="0" to-layer="899" to-port="1"/>
+ <edge from-layer="899" from-port="2" to-layer="901" to-port="0"/>
+ <edge from-layer="900" from-port="0" to-layer="901" to-port="1"/>
+ <edge from-layer="901" from-port="2" to-layer="902" to-port="0"/>
+ <edge from-layer="887" from-port="0" to-layer="902" to-port="1"/>
+ <edge from-layer="888" from-port="0" to-layer="902" to-port="2"/>
+ <edge from-layer="889" from-port="0" to-layer="902" to-port="3"/>
+ <edge from-layer="890" from-port="0" to-layer="902" to-port="4"/>
+ <edge from-layer="904" from-port="0" to-layer="905" to-port="0"/>
+ <edge from-layer="905" from-port="1" to-layer="907" to-port="0"/>
+ <edge from-layer="906" from-port="0" to-layer="907" to-port="1"/>
+ <edge from-layer="907" from-port="2" to-layer="909" to-port="0"/>
+ <edge from-layer="908" from-port="0" to-layer="909" to-port="1"/>
+ <edge from-layer="909" from-port="2" to-layer="910" to-port="0"/>
+ <edge from-layer="903" from-port="0" to-layer="910" to-port="1"/>
+ <edge from-layer="902" from-port="5" to-layer="911" to-port="0"/>
+ <edge from-layer="910" from-port="2" to-layer="911" to-port="1"/>
+ <edge from-layer="911" from-port="2" to-layer="913" to-port="0"/>
+ <edge from-layer="912" from-port="0" to-layer="913" to-port="1"/>
+ <edge from-layer="913" from-port="2" to-layer="915" to-port="0"/>
+ <edge from-layer="914" from-port="0" to-layer="915" to-port="1"/>
+ <edge from-layer="915" from-port="2" to-layer="916" to-port="0"/>
+ <edge from-layer="883" from-port="0" to-layer="916" to-port="1"/>
+ <edge from-layer="884" from-port="0" to-layer="916" to-port="2"/>
+ <edge from-layer="885" from-port="0" to-layer="916" to-port="3"/>
+ <edge from-layer="886" from-port="0" to-layer="916" to-port="4"/>
+ <edge from-layer="917" from-port="0" to-layer="918" to-port="0"/>
+ <edge from-layer="918" from-port="1" to-layer="920" to-port="0"/>
+ <edge from-layer="919" from-port="0" to-layer="920" to-port="1"/>
+ <edge from-layer="920" from-port="2" to-layer="922" to-port="0"/>
+ <edge from-layer="921" from-port="0" to-layer="922" to-port="1"/>
+ <edge from-layer="916" from-port="5" to-layer="923" to-port="0"/>
+ <edge from-layer="922" from-port="2" to-layer="923" to-port="1"/>
+ <edge from-layer="923" from-port="2" to-layer="925" to-port="0"/>
+ <edge from-layer="924" from-port="0" to-layer="925" to-port="1"/>
+ <edge from-layer="925" from-port="2" to-layer="926" to-port="0"/>
+ <edge from-layer="879" from-port="0" to-layer="926" to-port="1"/>
+ <edge from-layer="880" from-port="0" to-layer="926" to-port="2"/>
+ <edge from-layer="881" from-port="0" to-layer="926" to-port="3"/>
+ <edge from-layer="882" from-port="0" to-layer="926" to-port="4"/>
+ <edge from-layer="878" from-port="5" to-layer="927" to-port="0"/>
+ <edge from-layer="926" from-port="5" to-layer="927" to-port="1"/>
+ <edge from-layer="927" from-port="2" to-layer="929" to-port="0"/>
+ <edge from-layer="928" from-port="0" to-layer="929" to-port="1"/>
+ <edge from-layer="929" from-port="2" to-layer="930" to-port="0"/>
+ <edge from-layer="92" from-port="0" to-layer="930" to-port="1"/>
+ <edge from-layer="93" from-port="0" to-layer="930" to-port="2"/>
+ <edge from-layer="94" from-port="0" to-layer="930" to-port="3"/>
+ <edge from-layer="95" from-port="0" to-layer="930" to-port="4"/>
+ <edge from-layer="943" from-port="0" to-layer="944" to-port="0"/>
+ <edge from-layer="944" from-port="1" to-layer="946" to-port="0"/>
+ <edge from-layer="945" from-port="0" to-layer="946" to-port="1"/>
+ <edge from-layer="946" from-port="2" to-layer="948" to-port="0"/>
+ <edge from-layer="947" from-port="0" to-layer="948" to-port="1"/>
+ <edge from-layer="930" from-port="5" to-layer="949" to-port="0"/>
+ <edge from-layer="948" from-port="2" to-layer="949" to-port="1"/>
+ <edge from-layer="949" from-port="2" to-layer="951" to-port="0"/>
+ <edge from-layer="950" from-port="0" to-layer="951" to-port="1"/>
+ <edge from-layer="951" from-port="2" to-layer="953" to-port="0"/>
+ <edge from-layer="952" from-port="0" to-layer="953" to-port="1"/>
+ <edge from-layer="953" from-port="2" to-layer="954" to-port="0"/>
+ <edge from-layer="939" from-port="0" to-layer="954" to-port="1"/>
+ <edge from-layer="940" from-port="0" to-layer="954" to-port="2"/>
+ <edge from-layer="941" from-port="0" to-layer="954" to-port="3"/>
+ <edge from-layer="942" from-port="0" to-layer="954" to-port="4"/>
+ <edge from-layer="956" from-port="0" to-layer="957" to-port="0"/>
+ <edge from-layer="957" from-port="1" to-layer="959" to-port="0"/>
+ <edge from-layer="958" from-port="0" to-layer="959" to-port="1"/>
+ <edge from-layer="959" from-port="2" to-layer="961" to-port="0"/>
+ <edge from-layer="960" from-port="0" to-layer="961" to-port="1"/>
+ <edge from-layer="961" from-port="2" to-layer="962" to-port="0"/>
+ <edge from-layer="955" from-port="0" to-layer="962" to-port="1"/>
+ <edge from-layer="954" from-port="5" to-layer="963" to-port="0"/>
+ <edge from-layer="962" from-port="2" to-layer="963" to-port="1"/>
+ <edge from-layer="963" from-port="2" to-layer="965" to-port="0"/>
+ <edge from-layer="964" from-port="0" to-layer="965" to-port="1"/>
+ <edge from-layer="965" from-port="2" to-layer="967" to-port="0"/>
+ <edge from-layer="966" from-port="0" to-layer="967" to-port="1"/>
+ <edge from-layer="967" from-port="2" to-layer="968" to-port="0"/>
+ <edge from-layer="935" from-port="0" to-layer="968" to-port="1"/>
+ <edge from-layer="936" from-port="0" to-layer="968" to-port="2"/>
+ <edge from-layer="937" from-port="0" to-layer="968" to-port="3"/>
+ <edge from-layer="938" from-port="0" to-layer="968" to-port="4"/>
+ <edge from-layer="969" from-port="0" to-layer="970" to-port="0"/>
+ <edge from-layer="970" from-port="1" to-layer="972" to-port="0"/>
+ <edge from-layer="971" from-port="0" to-layer="972" to-port="1"/>
+ <edge from-layer="972" from-port="2" to-layer="974" to-port="0"/>
+ <edge from-layer="973" from-port="0" to-layer="974" to-port="1"/>
+ <edge from-layer="968" from-port="5" to-layer="975" to-port="0"/>
+ <edge from-layer="974" from-port="2" to-layer="975" to-port="1"/>
+ <edge from-layer="975" from-port="2" to-layer="977" to-port="0"/>
+ <edge from-layer="976" from-port="0" to-layer="977" to-port="1"/>
+ <edge from-layer="977" from-port="2" to-layer="978" to-port="0"/>
+ <edge from-layer="931" from-port="0" to-layer="978" to-port="1"/>
+ <edge from-layer="932" from-port="0" to-layer="978" to-port="2"/>
+ <edge from-layer="933" from-port="0" to-layer="978" to-port="3"/>
+ <edge from-layer="934" from-port="0" to-layer="978" to-port="4"/>
+ <edge from-layer="930" from-port="5" to-layer="979" to-port="0"/>
+ <edge from-layer="978" from-port="5" to-layer="979" to-port="1"/>
+ <edge from-layer="979" from-port="2" to-layer="981" to-port="0"/>
+ <edge from-layer="980" from-port="0" to-layer="981" to-port="1"/>
+ <edge from-layer="981" from-port="2" to-layer="982" to-port="0"/>
+ <edge from-layer="88" from-port="0" to-layer="982" to-port="1"/>
+ <edge from-layer="89" from-port="0" to-layer="982" to-port="2"/>
+ <edge from-layer="90" from-port="0" to-layer="982" to-port="3"/>
+ <edge from-layer="91" from-port="0" to-layer="982" to-port="4"/>
+ <edge from-layer="995" from-port="0" to-layer="996" to-port="0"/>
+ <edge from-layer="996" from-port="1" to-layer="998" to-port="0"/>
+ <edge from-layer="997" from-port="0" to-layer="998" to-port="1"/>
+ <edge from-layer="998" from-port="2" to-layer="1000" to-port="0"/>
+ <edge from-layer="999" from-port="0" to-layer="1000" to-port="1"/>
+ <edge from-layer="982" from-port="5" to-layer="1001" to-port="0"/>
+ <edge from-layer="1000" from-port="2" to-layer="1001" to-port="1"/>
+ <edge from-layer="1001" from-port="2" to-layer="1003" to-port="0"/>
+ <edge from-layer="1002" from-port="0" to-layer="1003" to-port="1"/>
+ <edge from-layer="1003" from-port="2" to-layer="1005" to-port="0"/>
+ <edge from-layer="1004" from-port="0" to-layer="1005" to-port="1"/>
+ <edge from-layer="1005" from-port="2" to-layer="1006" to-port="0"/>
+ <edge from-layer="991" from-port="0" to-layer="1006" to-port="1"/>
+ <edge from-layer="992" from-port="0" to-layer="1006" to-port="2"/>
+ <edge from-layer="993" from-port="0" to-layer="1006" to-port="3"/>
+ <edge from-layer="994" from-port="0" to-layer="1006" to-port="4"/>
+ <edge from-layer="1008" from-port="0" to-layer="1009" to-port="0"/>
+ <edge from-layer="1009" from-port="1" to-layer="1011" to-port="0"/>
+ <edge from-layer="1010" from-port="0" to-layer="1011" to-port="1"/>
+ <edge from-layer="1011" from-port="2" to-layer="1013" to-port="0"/>
+ <edge from-layer="1012" from-port="0" to-layer="1013" to-port="1"/>
+ <edge from-layer="1013" from-port="2" to-layer="1014" to-port="0"/>
+ <edge from-layer="1007" from-port="0" to-layer="1014" to-port="1"/>
+ <edge from-layer="1006" from-port="5" to-layer="1015" to-port="0"/>
+ <edge from-layer="1014" from-port="2" to-layer="1015" to-port="1"/>
+ <edge from-layer="1015" from-port="2" to-layer="1017" to-port="0"/>
+ <edge from-layer="1016" from-port="0" to-layer="1017" to-port="1"/>
+ <edge from-layer="1017" from-port="2" to-layer="1019" to-port="0"/>
+ <edge from-layer="1018" from-port="0" to-layer="1019" to-port="1"/>
+ <edge from-layer="1019" from-port="2" to-layer="1020" to-port="0"/>
+ <edge from-layer="987" from-port="0" to-layer="1020" to-port="1"/>
+ <edge from-layer="988" from-port="0" to-layer="1020" to-port="2"/>
+ <edge from-layer="989" from-port="0" to-layer="1020" to-port="3"/>
+ <edge from-layer="990" from-port="0" to-layer="1020" to-port="4"/>
+ <edge from-layer="1021" from-port="0" to-layer="1022" to-port="0"/>
+ <edge from-layer="1022" from-port="1" to-layer="1024" to-port="0"/>
+ <edge from-layer="1023" from-port="0" to-layer="1024" to-port="1"/>
+ <edge from-layer="1024" from-port="2" to-layer="1026" to-port="0"/>
+ <edge from-layer="1025" from-port="0" to-layer="1026" to-port="1"/>
+ <edge from-layer="1020" from-port="5" to-layer="1027" to-port="0"/>
+ <edge from-layer="1026" from-port="2" to-layer="1027" to-port="1"/>
+ <edge from-layer="1027" from-port="2" to-layer="1029" to-port="0"/>
+ <edge from-layer="1028" from-port="0" to-layer="1029" to-port="1"/>
+ <edge from-layer="1029" from-port="2" to-layer="1030" to-port="0"/>
+ <edge from-layer="983" from-port="0" to-layer="1030" to-port="1"/>
+ <edge from-layer="984" from-port="0" to-layer="1030" to-port="2"/>
+ <edge from-layer="985" from-port="0" to-layer="1030" to-port="3"/>
+ <edge from-layer="986" from-port="0" to-layer="1030" to-port="4"/>
+ <edge from-layer="982" from-port="5" to-layer="1031" to-port="0"/>
+ <edge from-layer="1030" from-port="5" to-layer="1031" to-port="1"/>
+ <edge from-layer="1031" from-port="2" to-layer="1033" to-port="0"/>
+ <edge from-layer="1032" from-port="0" to-layer="1033" to-port="1"/>
+ <edge from-layer="1033" from-port="2" to-layer="1034" to-port="0"/>
+ <edge from-layer="84" from-port="0" to-layer="1034" to-port="1"/>
+ <edge from-layer="85" from-port="0" to-layer="1034" to-port="2"/>
+ <edge from-layer="86" from-port="0" to-layer="1034" to-port="3"/>
+ <edge from-layer="87" from-port="0" to-layer="1034" to-port="4"/>
+ <edge from-layer="1047" from-port="0" to-layer="1048" to-port="0"/>
+ <edge from-layer="1048" from-port="1" to-layer="1050" to-port="0"/>
+ <edge from-layer="1049" from-port="0" to-layer="1050" to-port="1"/>
+ <edge from-layer="1050" from-port="2" to-layer="1052" to-port="0"/>
+ <edge from-layer="1051" from-port="0" to-layer="1052" to-port="1"/>
+ <edge from-layer="1034" from-port="5" to-layer="1053" to-port="0"/>
+ <edge from-layer="1052" from-port="2" to-layer="1053" to-port="1"/>
+ <edge from-layer="1053" from-port="2" to-layer="1055" to-port="0"/>
+ <edge from-layer="1054" from-port="0" to-layer="1055" to-port="1"/>
+ <edge from-layer="1055" from-port="2" to-layer="1057" to-port="0"/>
+ <edge from-layer="1056" from-port="0" to-layer="1057" to-port="1"/>
+ <edge from-layer="1057" from-port="2" to-layer="1058" to-port="0"/>
+ <edge from-layer="1043" from-port="0" to-layer="1058" to-port="1"/>
+ <edge from-layer="1044" from-port="0" to-layer="1058" to-port="2"/>
+ <edge from-layer="1045" from-port="0" to-layer="1058" to-port="3"/>
+ <edge from-layer="1046" from-port="0" to-layer="1058" to-port="4"/>
+ <edge from-layer="1060" from-port="0" to-layer="1061" to-port="0"/>
+ <edge from-layer="1061" from-port="1" to-layer="1063" to-port="0"/>
+ <edge from-layer="1062" from-port="0" to-layer="1063" to-port="1"/>
+ <edge from-layer="1063" from-port="2" to-layer="1065" to-port="0"/>
+ <edge from-layer="1064" from-port="0" to-layer="1065" to-port="1"/>
+ <edge from-layer="1065" from-port="2" to-layer="1066" to-port="0"/>
+ <edge from-layer="1059" from-port="0" to-layer="1066" to-port="1"/>
+ <edge from-layer="1058" from-port="5" to-layer="1067" to-port="0"/>
+ <edge from-layer="1066" from-port="2" to-layer="1067" to-port="1"/>
+ <edge from-layer="1067" from-port="2" to-layer="1069" to-port="0"/>
+ <edge from-layer="1068" from-port="0" to-layer="1069" to-port="1"/>
+ <edge from-layer="1069" from-port="2" to-layer="1071" to-port="0"/>
+ <edge from-layer="1070" from-port="0" to-layer="1071" to-port="1"/>
+ <edge from-layer="1071" from-port="2" to-layer="1072" to-port="0"/>
+ <edge from-layer="1039" from-port="0" to-layer="1072" to-port="1"/>
+ <edge from-layer="1040" from-port="0" to-layer="1072" to-port="2"/>
+ <edge from-layer="1041" from-port="0" to-layer="1072" to-port="3"/>
+ <edge from-layer="1042" from-port="0" to-layer="1072" to-port="4"/>
+ <edge from-layer="1073" from-port="0" to-layer="1074" to-port="0"/>
+ <edge from-layer="1074" from-port="1" to-layer="1076" to-port="0"/>
+ <edge from-layer="1075" from-port="0" to-layer="1076" to-port="1"/>
+ <edge from-layer="1076" from-port="2" to-layer="1078" to-port="0"/>
+ <edge from-layer="1077" from-port="0" to-layer="1078" to-port="1"/>
+ <edge from-layer="1072" from-port="5" to-layer="1079" to-port="0"/>
+ <edge from-layer="1078" from-port="2" to-layer="1079" to-port="1"/>
+ <edge from-layer="1079" from-port="2" to-layer="1081" to-port="0"/>
+ <edge from-layer="1080" from-port="0" to-layer="1081" to-port="1"/>
+ <edge from-layer="1081" from-port="2" to-layer="1082" to-port="0"/>
+ <edge from-layer="1035" from-port="0" to-layer="1082" to-port="1"/>
+ <edge from-layer="1036" from-port="0" to-layer="1082" to-port="2"/>
+ <edge from-layer="1037" from-port="0" to-layer="1082" to-port="3"/>
+ <edge from-layer="1038" from-port="0" to-layer="1082" to-port="4"/>
+ <edge from-layer="1034" from-port="5" to-layer="1083" to-port="0"/>
+ <edge from-layer="1082" from-port="5" to-layer="1083" to-port="1"/>
+ <edge from-layer="1083" from-port="2" to-layer="1085" to-port="0"/>
+ <edge from-layer="1084" from-port="0" to-layer="1085" to-port="1"/>
+ <edge from-layer="1085" from-port="2" to-layer="1086" to-port="0"/>
+ <edge from-layer="80" from-port="0" to-layer="1086" to-port="1"/>
+ <edge from-layer="81" from-port="0" to-layer="1086" to-port="2"/>
+ <edge from-layer="82" from-port="0" to-layer="1086" to-port="3"/>
+ <edge from-layer="83" from-port="0" to-layer="1086" to-port="4"/>
+ <edge from-layer="1099" from-port="0" to-layer="1100" to-port="0"/>
+ <edge from-layer="1100" from-port="1" to-layer="1102" to-port="0"/>
+ <edge from-layer="1101" from-port="0" to-layer="1102" to-port="1"/>
+ <edge from-layer="1102" from-port="2" to-layer="1104" to-port="0"/>
+ <edge from-layer="1103" from-port="0" to-layer="1104" to-port="1"/>
+ <edge from-layer="1086" from-port="5" to-layer="1105" to-port="0"/>
+ <edge from-layer="1104" from-port="2" to-layer="1105" to-port="1"/>
+ <edge from-layer="1105" from-port="2" to-layer="1107" to-port="0"/>
+ <edge from-layer="1106" from-port="0" to-layer="1107" to-port="1"/>
+ <edge from-layer="1107" from-port="2" to-layer="1109" to-port="0"/>
+ <edge from-layer="1108" from-port="0" to-layer="1109" to-port="1"/>
+ <edge from-layer="1109" from-port="2" to-layer="1110" to-port="0"/>
+ <edge from-layer="1095" from-port="0" to-layer="1110" to-port="1"/>
+ <edge from-layer="1096" from-port="0" to-layer="1110" to-port="2"/>
+ <edge from-layer="1097" from-port="0" to-layer="1110" to-port="3"/>
+ <edge from-layer="1098" from-port="0" to-layer="1110" to-port="4"/>
+ <edge from-layer="1112" from-port="0" to-layer="1113" to-port="0"/>
+ <edge from-layer="1113" from-port="1" to-layer="1115" to-port="0"/>
+ <edge from-layer="1114" from-port="0" to-layer="1115" to-port="1"/>
+ <edge from-layer="1115" from-port="2" to-layer="1117" to-port="0"/>
+ <edge from-layer="1116" from-port="0" to-layer="1117" to-port="1"/>
+ <edge from-layer="1117" from-port="2" to-layer="1118" to-port="0"/>
+ <edge from-layer="1111" from-port="0" to-layer="1118" to-port="1"/>
+ <edge from-layer="1110" from-port="5" to-layer="1119" to-port="0"/>
+ <edge from-layer="1118" from-port="2" to-layer="1119" to-port="1"/>
+ <edge from-layer="1119" from-port="2" to-layer="1121" to-port="0"/>
+ <edge from-layer="1120" from-port="0" to-layer="1121" to-port="1"/>
+ <edge from-layer="1121" from-port="2" to-layer="1123" to-port="0"/>
+ <edge from-layer="1122" from-port="0" to-layer="1123" to-port="1"/>
+ <edge from-layer="1123" from-port="2" to-layer="1124" to-port="0"/>
+ <edge from-layer="1091" from-port="0" to-layer="1124" to-port="1"/>
+ <edge from-layer="1092" from-port="0" to-layer="1124" to-port="2"/>
+ <edge from-layer="1093" from-port="0" to-layer="1124" to-port="3"/>
+ <edge from-layer="1094" from-port="0" to-layer="1124" to-port="4"/>
+ <edge from-layer="1125" from-port="0" to-layer="1126" to-port="0"/>
+ <edge from-layer="1126" from-port="1" to-layer="1128" to-port="0"/>
+ <edge from-layer="1127" from-port="0" to-layer="1128" to-port="1"/>
+ <edge from-layer="1128" from-port="2" to-layer="1130" to-port="0"/>
+ <edge from-layer="1129" from-port="0" to-layer="1130" to-port="1"/>
+ <edge from-layer="1124" from-port="5" to-layer="1131" to-port="0"/>
+ <edge from-layer="1130" from-port="2" to-layer="1131" to-port="1"/>
+ <edge from-layer="1131" from-port="2" to-layer="1133" to-port="0"/>
+ <edge from-layer="1132" from-port="0" to-layer="1133" to-port="1"/>
+ <edge from-layer="1133" from-port="2" to-layer="1134" to-port="0"/>
+ <edge from-layer="1087" from-port="0" to-layer="1134" to-port="1"/>
+ <edge from-layer="1088" from-port="0" to-layer="1134" to-port="2"/>
+ <edge from-layer="1089" from-port="0" to-layer="1134" to-port="3"/>
+ <edge from-layer="1090" from-port="0" to-layer="1134" to-port="4"/>
+ <edge from-layer="1086" from-port="5" to-layer="1135" to-port="0"/>
+ <edge from-layer="1134" from-port="5" to-layer="1135" to-port="1"/>
+ <edge from-layer="1135" from-port="2" to-layer="1137" to-port="0"/>
+ <edge from-layer="1136" from-port="0" to-layer="1137" to-port="1"/>
+ <edge from-layer="1137" from-port="2" to-layer="1138" to-port="0"/>
+ <edge from-layer="76" from-port="0" to-layer="1138" to-port="1"/>
+ <edge from-layer="77" from-port="0" to-layer="1138" to-port="2"/>
+ <edge from-layer="78" from-port="0" to-layer="1138" to-port="3"/>
+ <edge from-layer="79" from-port="0" to-layer="1138" to-port="4"/>
+ <edge from-layer="1151" from-port="0" to-layer="1152" to-port="0"/>
+ <edge from-layer="1152" from-port="1" to-layer="1154" to-port="0"/>
+ <edge from-layer="1153" from-port="0" to-layer="1154" to-port="1"/>
+ <edge from-layer="1154" from-port="2" to-layer="1156" to-port="0"/>
+ <edge from-layer="1155" from-port="0" to-layer="1156" to-port="1"/>
+ <edge from-layer="1138" from-port="5" to-layer="1157" to-port="0"/>
+ <edge from-layer="1156" from-port="2" to-layer="1157" to-port="1"/>
+ <edge from-layer="1157" from-port="2" to-layer="1159" to-port="0"/>
+ <edge from-layer="1158" from-port="0" to-layer="1159" to-port="1"/>
+ <edge from-layer="1159" from-port="2" to-layer="1161" to-port="0"/>
+ <edge from-layer="1160" from-port="0" to-layer="1161" to-port="1"/>
+ <edge from-layer="1161" from-port="2" to-layer="1162" to-port="0"/>
+ <edge from-layer="1147" from-port="0" to-layer="1162" to-port="1"/>
+ <edge from-layer="1148" from-port="0" to-layer="1162" to-port="2"/>
+ <edge from-layer="1149" from-port="0" to-layer="1162" to-port="3"/>
+ <edge from-layer="1150" from-port="0" to-layer="1162" to-port="4"/>
+ <edge from-layer="1164" from-port="0" to-layer="1165" to-port="0"/>
+ <edge from-layer="1165" from-port="1" to-layer="1167" to-port="0"/>
+ <edge from-layer="1166" from-port="0" to-layer="1167" to-port="1"/>
+ <edge from-layer="1167" from-port="2" to-layer="1169" to-port="0"/>
+ <edge from-layer="1168" from-port="0" to-layer="1169" to-port="1"/>
+ <edge from-layer="1169" from-port="2" to-layer="1170" to-port="0"/>
+ <edge from-layer="1163" from-port="0" to-layer="1170" to-port="1"/>
+ <edge from-layer="1162" from-port="5" to-layer="1171" to-port="0"/>
+ <edge from-layer="1170" from-port="2" to-layer="1171" to-port="1"/>
+ <edge from-layer="1171" from-port="2" to-layer="1173" to-port="0"/>
+ <edge from-layer="1172" from-port="0" to-layer="1173" to-port="1"/>
+ <edge from-layer="1173" from-port="2" to-layer="1175" to-port="0"/>
+ <edge from-layer="1174" from-port="0" to-layer="1175" to-port="1"/>
+ <edge from-layer="1175" from-port="2" to-layer="1176" to-port="0"/>
+ <edge from-layer="1143" from-port="0" to-layer="1176" to-port="1"/>
+ <edge from-layer="1144" from-port="0" to-layer="1176" to-port="2"/>
+ <edge from-layer="1145" from-port="0" to-layer="1176" to-port="3"/>
+ <edge from-layer="1146" from-port="0" to-layer="1176" to-port="4"/>
+ <edge from-layer="1177" from-port="0" to-layer="1178" to-port="0"/>
+ <edge from-layer="1178" from-port="1" to-layer="1180" to-port="0"/>
+ <edge from-layer="1179" from-port="0" to-layer="1180" to-port="1"/>
+ <edge from-layer="1180" from-port="2" to-layer="1182" to-port="0"/>
+ <edge from-layer="1181" from-port="0" to-layer="1182" to-port="1"/>
+ <edge from-layer="1176" from-port="5" to-layer="1183" to-port="0"/>
+ <edge from-layer="1182" from-port="2" to-layer="1183" to-port="1"/>
+ <edge from-layer="1183" from-port="2" to-layer="1185" to-port="0"/>
+ <edge from-layer="1184" from-port="0" to-layer="1185" to-port="1"/>
+ <edge from-layer="1185" from-port="2" to-layer="1186" to-port="0"/>
+ <edge from-layer="1139" from-port="0" to-layer="1186" to-port="1"/>
+ <edge from-layer="1140" from-port="0" to-layer="1186" to-port="2"/>
+ <edge from-layer="1141" from-port="0" to-layer="1186" to-port="3"/>
+ <edge from-layer="1142" from-port="0" to-layer="1186" to-port="4"/>
+ <edge from-layer="1138" from-port="5" to-layer="1187" to-port="0"/>
+ <edge from-layer="1186" from-port="5" to-layer="1187" to-port="1"/>
+ <edge from-layer="1187" from-port="2" to-layer="1189" to-port="0"/>
+ <edge from-layer="1188" from-port="0" to-layer="1189" to-port="1"/>
+ <edge from-layer="1189" from-port="2" to-layer="1190" to-port="0"/>
+ <edge from-layer="72" from-port="0" to-layer="1190" to-port="1"/>
+ <edge from-layer="73" from-port="0" to-layer="1190" to-port="2"/>
+ <edge from-layer="74" from-port="0" to-layer="1190" to-port="3"/>
+ <edge from-layer="75" from-port="0" to-layer="1190" to-port="4"/>
+ <edge from-layer="1203" from-port="0" to-layer="1204" to-port="0"/>
+ <edge from-layer="1204" from-port="1" to-layer="1206" to-port="0"/>
+ <edge from-layer="1205" from-port="0" to-layer="1206" to-port="1"/>
+ <edge from-layer="1206" from-port="2" to-layer="1208" to-port="0"/>
+ <edge from-layer="1207" from-port="0" to-layer="1208" to-port="1"/>
+ <edge from-layer="1190" from-port="5" to-layer="1209" to-port="0"/>
+ <edge from-layer="1208" from-port="2" to-layer="1209" to-port="1"/>
+ <edge from-layer="1209" from-port="2" to-layer="1211" to-port="0"/>
+ <edge from-layer="1210" from-port="0" to-layer="1211" to-port="1"/>
+ <edge from-layer="1211" from-port="2" to-layer="1213" to-port="0"/>
+ <edge from-layer="1212" from-port="0" to-layer="1213" to-port="1"/>
+ <edge from-layer="1213" from-port="2" to-layer="1214" to-port="0"/>
+ <edge from-layer="1199" from-port="0" to-layer="1214" to-port="1"/>
+ <edge from-layer="1200" from-port="0" to-layer="1214" to-port="2"/>
+ <edge from-layer="1201" from-port="0" to-layer="1214" to-port="3"/>
+ <edge from-layer="1202" from-port="0" to-layer="1214" to-port="4"/>
+ <edge from-layer="1216" from-port="0" to-layer="1217" to-port="0"/>
+ <edge from-layer="1217" from-port="1" to-layer="1219" to-port="0"/>
+ <edge from-layer="1218" from-port="0" to-layer="1219" to-port="1"/>
+ <edge from-layer="1219" from-port="2" to-layer="1221" to-port="0"/>
+ <edge from-layer="1220" from-port="0" to-layer="1221" to-port="1"/>
+ <edge from-layer="1221" from-port="2" to-layer="1222" to-port="0"/>
+ <edge from-layer="1215" from-port="0" to-layer="1222" to-port="1"/>
+ <edge from-layer="1214" from-port="5" to-layer="1223" to-port="0"/>
+ <edge from-layer="1222" from-port="2" to-layer="1223" to-port="1"/>
+ <edge from-layer="1223" from-port="2" to-layer="1225" to-port="0"/>
+ <edge from-layer="1224" from-port="0" to-layer="1225" to-port="1"/>
+ <edge from-layer="1225" from-port="2" to-layer="1227" to-port="0"/>
+ <edge from-layer="1226" from-port="0" to-layer="1227" to-port="1"/>
+ <edge from-layer="1227" from-port="2" to-layer="1228" to-port="0"/>
+ <edge from-layer="1195" from-port="0" to-layer="1228" to-port="1"/>
+ <edge from-layer="1196" from-port="0" to-layer="1228" to-port="2"/>
+ <edge from-layer="1197" from-port="0" to-layer="1228" to-port="3"/>
+ <edge from-layer="1198" from-port="0" to-layer="1228" to-port="4"/>
+ <edge from-layer="1229" from-port="0" to-layer="1230" to-port="0"/>
+ <edge from-layer="1230" from-port="1" to-layer="1232" to-port="0"/>
+ <edge from-layer="1231" from-port="0" to-layer="1232" to-port="1"/>
+ <edge from-layer="1232" from-port="2" to-layer="1234" to-port="0"/>
+ <edge from-layer="1233" from-port="0" to-layer="1234" to-port="1"/>
+ <edge from-layer="1228" from-port="5" to-layer="1235" to-port="0"/>
+ <edge from-layer="1234" from-port="2" to-layer="1235" to-port="1"/>
+ <edge from-layer="1235" from-port="2" to-layer="1237" to-port="0"/>
+ <edge from-layer="1236" from-port="0" to-layer="1237" to-port="1"/>
+ <edge from-layer="1237" from-port="2" to-layer="1238" to-port="0"/>
+ <edge from-layer="1191" from-port="0" to-layer="1238" to-port="1"/>
+ <edge from-layer="1192" from-port="0" to-layer="1238" to-port="2"/>
+ <edge from-layer="1193" from-port="0" to-layer="1238" to-port="3"/>
+ <edge from-layer="1194" from-port="0" to-layer="1238" to-port="4"/>
+ <edge from-layer="1190" from-port="5" to-layer="1239" to-port="0"/>
+ <edge from-layer="1238" from-port="5" to-layer="1239" to-port="1"/>
+ <edge from-layer="1239" from-port="2" to-layer="1241" to-port="0"/>
+ <edge from-layer="1240" from-port="0" to-layer="1241" to-port="1"/>
+ <edge from-layer="1241" from-port="2" to-layer="1242" to-port="0"/>
+ <edge from-layer="68" from-port="0" to-layer="1242" to-port="1"/>
+ <edge from-layer="69" from-port="0" to-layer="1242" to-port="2"/>
+ <edge from-layer="70" from-port="0" to-layer="1242" to-port="3"/>
+ <edge from-layer="71" from-port="0" to-layer="1242" to-port="4"/>
+ <edge from-layer="1255" from-port="0" to-layer="1256" to-port="0"/>
+ <edge from-layer="1256" from-port="1" to-layer="1258" to-port="0"/>
+ <edge from-layer="1257" from-port="0" to-layer="1258" to-port="1"/>
+ <edge from-layer="1258" from-port="2" to-layer="1260" to-port="0"/>
+ <edge from-layer="1259" from-port="0" to-layer="1260" to-port="1"/>
+ <edge from-layer="1242" from-port="5" to-layer="1261" to-port="0"/>
+ <edge from-layer="1260" from-port="2" to-layer="1261" to-port="1"/>
+ <edge from-layer="1261" from-port="2" to-layer="1263" to-port="0"/>
+ <edge from-layer="1262" from-port="0" to-layer="1263" to-port="1"/>
+ <edge from-layer="1263" from-port="2" to-layer="1265" to-port="0"/>
+ <edge from-layer="1264" from-port="0" to-layer="1265" to-port="1"/>
+ <edge from-layer="1265" from-port="2" to-layer="1266" to-port="0"/>
+ <edge from-layer="1251" from-port="0" to-layer="1266" to-port="1"/>
+ <edge from-layer="1252" from-port="0" to-layer="1266" to-port="2"/>
+ <edge from-layer="1253" from-port="0" to-layer="1266" to-port="3"/>
+ <edge from-layer="1254" from-port="0" to-layer="1266" to-port="4"/>
+ <edge from-layer="1268" from-port="0" to-layer="1269" to-port="0"/>
+ <edge from-layer="1269" from-port="1" to-layer="1271" to-port="0"/>
+ <edge from-layer="1270" from-port="0" to-layer="1271" to-port="1"/>
+ <edge from-layer="1271" from-port="2" to-layer="1273" to-port="0"/>
+ <edge from-layer="1272" from-port="0" to-layer="1273" to-port="1"/>
+ <edge from-layer="1273" from-port="2" to-layer="1274" to-port="0"/>
+ <edge from-layer="1267" from-port="0" to-layer="1274" to-port="1"/>
+ <edge from-layer="1266" from-port="5" to-layer="1275" to-port="0"/>
+ <edge from-layer="1274" from-port="2" to-layer="1275" to-port="1"/>
+ <edge from-layer="1275" from-port="2" to-layer="1277" to-port="0"/>
+ <edge from-layer="1276" from-port="0" to-layer="1277" to-port="1"/>
+ <edge from-layer="1277" from-port="2" to-layer="1279" to-port="0"/>
+ <edge from-layer="1278" from-port="0" to-layer="1279" to-port="1"/>
+ <edge from-layer="1279" from-port="2" to-layer="1280" to-port="0"/>
+ <edge from-layer="1247" from-port="0" to-layer="1280" to-port="1"/>
+ <edge from-layer="1248" from-port="0" to-layer="1280" to-port="2"/>
+ <edge from-layer="1249" from-port="0" to-layer="1280" to-port="3"/>
+ <edge from-layer="1250" from-port="0" to-layer="1280" to-port="4"/>
+ <edge from-layer="1281" from-port="0" to-layer="1282" to-port="0"/>
+ <edge from-layer="1282" from-port="1" to-layer="1284" to-port="0"/>
+ <edge from-layer="1283" from-port="0" to-layer="1284" to-port="1"/>
+ <edge from-layer="1284" from-port="2" to-layer="1286" to-port="0"/>
+ <edge from-layer="1285" from-port="0" to-layer="1286" to-port="1"/>
+ <edge from-layer="1280" from-port="5" to-layer="1287" to-port="0"/>
+ <edge from-layer="1286" from-port="2" to-layer="1287" to-port="1"/>
+ <edge from-layer="1287" from-port="2" to-layer="1289" to-port="0"/>
+ <edge from-layer="1288" from-port="0" to-layer="1289" to-port="1"/>
+ <edge from-layer="1289" from-port="2" to-layer="1290" to-port="0"/>
+ <edge from-layer="1243" from-port="0" to-layer="1290" to-port="1"/>
+ <edge from-layer="1244" from-port="0" to-layer="1290" to-port="2"/>
+ <edge from-layer="1245" from-port="0" to-layer="1290" to-port="3"/>
+ <edge from-layer="1246" from-port="0" to-layer="1290" to-port="4"/>
+ <edge from-layer="1242" from-port="5" to-layer="1291" to-port="0"/>
+ <edge from-layer="1290" from-port="5" to-layer="1291" to-port="1"/>
+ <edge from-layer="1291" from-port="2" to-layer="1293" to-port="0"/>
+ <edge from-layer="1292" from-port="0" to-layer="1293" to-port="1"/>
+ <edge from-layer="1293" from-port="2" to-layer="1294" to-port="0"/>
+ <edge from-layer="64" from-port="0" to-layer="1294" to-port="1"/>
+ <edge from-layer="65" from-port="0" to-layer="1294" to-port="2"/>
+ <edge from-layer="66" from-port="0" to-layer="1294" to-port="3"/>
+ <edge from-layer="67" from-port="0" to-layer="1294" to-port="4"/>
+ <edge from-layer="1307" from-port="0" to-layer="1308" to-port="0"/>
+ <edge from-layer="1308" from-port="1" to-layer="1310" to-port="0"/>
+ <edge from-layer="1309" from-port="0" to-layer="1310" to-port="1"/>
+ <edge from-layer="1310" from-port="2" to-layer="1312" to-port="0"/>
+ <edge from-layer="1311" from-port="0" to-layer="1312" to-port="1"/>
+ <edge from-layer="1294" from-port="5" to-layer="1313" to-port="0"/>
+ <edge from-layer="1312" from-port="2" to-layer="1313" to-port="1"/>
+ <edge from-layer="1313" from-port="2" to-layer="1315" to-port="0"/>
+ <edge from-layer="1314" from-port="0" to-layer="1315" to-port="1"/>
+ <edge from-layer="1315" from-port="2" to-layer="1317" to-port="0"/>
+ <edge from-layer="1316" from-port="0" to-layer="1317" to-port="1"/>
+ <edge from-layer="1317" from-port="2" to-layer="1318" to-port="0"/>
+ <edge from-layer="1303" from-port="0" to-layer="1318" to-port="1"/>
+ <edge from-layer="1304" from-port="0" to-layer="1318" to-port="2"/>
+ <edge from-layer="1305" from-port="0" to-layer="1318" to-port="3"/>
+ <edge from-layer="1306" from-port="0" to-layer="1318" to-port="4"/>
+ <edge from-layer="1320" from-port="0" to-layer="1321" to-port="0"/>
+ <edge from-layer="1321" from-port="1" to-layer="1323" to-port="0"/>
+ <edge from-layer="1322" from-port="0" to-layer="1323" to-port="1"/>
+ <edge from-layer="1323" from-port="2" to-layer="1325" to-port="0"/>
+ <edge from-layer="1324" from-port="0" to-layer="1325" to-port="1"/>
+ <edge from-layer="1325" from-port="2" to-layer="1326" to-port="0"/>
+ <edge from-layer="1319" from-port="0" to-layer="1326" to-port="1"/>
+ <edge from-layer="1318" from-port="5" to-layer="1327" to-port="0"/>
+ <edge from-layer="1326" from-port="2" to-layer="1327" to-port="1"/>
+ <edge from-layer="1327" from-port="2" to-layer="1329" to-port="0"/>
+ <edge from-layer="1328" from-port="0" to-layer="1329" to-port="1"/>
+ <edge from-layer="1329" from-port="2" to-layer="1331" to-port="0"/>
+ <edge from-layer="1330" from-port="0" to-layer="1331" to-port="1"/>
+ <edge from-layer="1331" from-port="2" to-layer="1332" to-port="0"/>
+ <edge from-layer="1299" from-port="0" to-layer="1332" to-port="1"/>
+ <edge from-layer="1300" from-port="0" to-layer="1332" to-port="2"/>
+ <edge from-layer="1301" from-port="0" to-layer="1332" to-port="3"/>
+ <edge from-layer="1302" from-port="0" to-layer="1332" to-port="4"/>
+ <edge from-layer="1333" from-port="0" to-layer="1334" to-port="0"/>
+ <edge from-layer="1334" from-port="1" to-layer="1336" to-port="0"/>
+ <edge from-layer="1335" from-port="0" to-layer="1336" to-port="1"/>
+ <edge from-layer="1336" from-port="2" to-layer="1338" to-port="0"/>
+ <edge from-layer="1337" from-port="0" to-layer="1338" to-port="1"/>
+ <edge from-layer="1332" from-port="5" to-layer="1339" to-port="0"/>
+ <edge from-layer="1338" from-port="2" to-layer="1339" to-port="1"/>
+ <edge from-layer="1339" from-port="2" to-layer="1341" to-port="0"/>
+ <edge from-layer="1340" from-port="0" to-layer="1341" to-port="1"/>
+ <edge from-layer="1341" from-port="2" to-layer="1342" to-port="0"/>
+ <edge from-layer="1295" from-port="0" to-layer="1342" to-port="1"/>
+ <edge from-layer="1296" from-port="0" to-layer="1342" to-port="2"/>
+ <edge from-layer="1297" from-port="0" to-layer="1342" to-port="3"/>
+ <edge from-layer="1298" from-port="0" to-layer="1342" to-port="4"/>
+ <edge from-layer="1294" from-port="5" to-layer="1343" to-port="0"/>
+ <edge from-layer="1342" from-port="5" to-layer="1343" to-port="1"/>
+ <edge from-layer="1343" from-port="2" to-layer="1345" to-port="0"/>
+ <edge from-layer="1344" from-port="0" to-layer="1345" to-port="1"/>
+ <edge from-layer="1345" from-port="2" to-layer="1346" to-port="0"/>
+ <edge from-layer="60" from-port="0" to-layer="1346" to-port="1"/>
+ <edge from-layer="61" from-port="0" to-layer="1346" to-port="2"/>
+ <edge from-layer="62" from-port="0" to-layer="1346" to-port="3"/>
+ <edge from-layer="63" from-port="0" to-layer="1346" to-port="4"/>
+ <edge from-layer="1359" from-port="0" to-layer="1360" to-port="0"/>
+ <edge from-layer="1360" from-port="1" to-layer="1362" to-port="0"/>
+ <edge from-layer="1361" from-port="0" to-layer="1362" to-port="1"/>
+ <edge from-layer="1362" from-port="2" to-layer="1364" to-port="0"/>
+ <edge from-layer="1363" from-port="0" to-layer="1364" to-port="1"/>
+ <edge from-layer="1346" from-port="5" to-layer="1365" to-port="0"/>
+ <edge from-layer="1364" from-port="2" to-layer="1365" to-port="1"/>
+ <edge from-layer="1365" from-port="2" to-layer="1367" to-port="0"/>
+ <edge from-layer="1366" from-port="0" to-layer="1367" to-port="1"/>
+ <edge from-layer="1367" from-port="2" to-layer="1369" to-port="0"/>
+ <edge from-layer="1368" from-port="0" to-layer="1369" to-port="1"/>
+ <edge from-layer="1369" from-port="2" to-layer="1370" to-port="0"/>
+ <edge from-layer="1355" from-port="0" to-layer="1370" to-port="1"/>
+ <edge from-layer="1356" from-port="0" to-layer="1370" to-port="2"/>
+ <edge from-layer="1357" from-port="0" to-layer="1370" to-port="3"/>
+ <edge from-layer="1358" from-port="0" to-layer="1370" to-port="4"/>
+ <edge from-layer="1372" from-port="0" to-layer="1373" to-port="0"/>
+ <edge from-layer="1373" from-port="1" to-layer="1375" to-port="0"/>
+ <edge from-layer="1374" from-port="0" to-layer="1375" to-port="1"/>
+ <edge from-layer="1375" from-port="2" to-layer="1377" to-port="0"/>
+ <edge from-layer="1376" from-port="0" to-layer="1377" to-port="1"/>
+ <edge from-layer="1377" from-port="2" to-layer="1378" to-port="0"/>
+ <edge from-layer="1371" from-port="0" to-layer="1378" to-port="1"/>
+ <edge from-layer="1370" from-port="5" to-layer="1379" to-port="0"/>
+ <edge from-layer="1378" from-port="2" to-layer="1379" to-port="1"/>
+ <edge from-layer="1379" from-port="2" to-layer="1381" to-port="0"/>
+ <edge from-layer="1380" from-port="0" to-layer="1381" to-port="1"/>
+ <edge from-layer="1381" from-port="2" to-layer="1383" to-port="0"/>
+ <edge from-layer="1382" from-port="0" to-layer="1383" to-port="1"/>
+ <edge from-layer="1383" from-port="2" to-layer="1384" to-port="0"/>
+ <edge from-layer="1351" from-port="0" to-layer="1384" to-port="1"/>
+ <edge from-layer="1352" from-port="0" to-layer="1384" to-port="2"/>
+ <edge from-layer="1353" from-port="0" to-layer="1384" to-port="3"/>
+ <edge from-layer="1354" from-port="0" to-layer="1384" to-port="4"/>
+ <edge from-layer="1385" from-port="0" to-layer="1386" to-port="0"/>
+ <edge from-layer="1386" from-port="1" to-layer="1388" to-port="0"/>
+ <edge from-layer="1387" from-port="0" to-layer="1388" to-port="1"/>
+ <edge from-layer="1388" from-port="2" to-layer="1390" to-port="0"/>
+ <edge from-layer="1389" from-port="0" to-layer="1390" to-port="1"/>
+ <edge from-layer="1384" from-port="5" to-layer="1391" to-port="0"/>
+ <edge from-layer="1390" from-port="2" to-layer="1391" to-port="1"/>
+ <edge from-layer="1391" from-port="2" to-layer="1393" to-port="0"/>
+ <edge from-layer="1392" from-port="0" to-layer="1393" to-port="1"/>
+ <edge from-layer="1393" from-port="2" to-layer="1394" to-port="0"/>
+ <edge from-layer="1347" from-port="0" to-layer="1394" to-port="1"/>
+ <edge from-layer="1348" from-port="0" to-layer="1394" to-port="2"/>
+ <edge from-layer="1349" from-port="0" to-layer="1394" to-port="3"/>
+ <edge from-layer="1350" from-port="0" to-layer="1394" to-port="4"/>
+ <edge from-layer="1346" from-port="5" to-layer="1395" to-port="0"/>
+ <edge from-layer="1394" from-port="5" to-layer="1395" to-port="1"/>
+ <edge from-layer="1395" from-port="2" to-layer="1397" to-port="0"/>
+ <edge from-layer="1396" from-port="0" to-layer="1397" to-port="1"/>
+ <edge from-layer="1397" from-port="2" to-layer="1398" to-port="0"/>
+ <edge from-layer="56" from-port="0" to-layer="1398" to-port="1"/>
+ <edge from-layer="57" from-port="0" to-layer="1398" to-port="2"/>
+ <edge from-layer="58" from-port="0" to-layer="1398" to-port="3"/>
+ <edge from-layer="59" from-port="0" to-layer="1398" to-port="4"/>
+ <edge from-layer="1411" from-port="0" to-layer="1412" to-port="0"/>
+ <edge from-layer="1412" from-port="1" to-layer="1414" to-port="0"/>
+ <edge from-layer="1413" from-port="0" to-layer="1414" to-port="1"/>
+ <edge from-layer="1414" from-port="2" to-layer="1416" to-port="0"/>
+ <edge from-layer="1415" from-port="0" to-layer="1416" to-port="1"/>
+ <edge from-layer="1398" from-port="5" to-layer="1417" to-port="0"/>
+ <edge from-layer="1416" from-port="2" to-layer="1417" to-port="1"/>
+ <edge from-layer="1417" from-port="2" to-layer="1419" to-port="0"/>
+ <edge from-layer="1418" from-port="0" to-layer="1419" to-port="1"/>
+ <edge from-layer="1419" from-port="2" to-layer="1421" to-port="0"/>
+ <edge from-layer="1420" from-port="0" to-layer="1421" to-port="1"/>
+ <edge from-layer="1421" from-port="2" to-layer="1422" to-port="0"/>
+ <edge from-layer="1407" from-port="0" to-layer="1422" to-port="1"/>
+ <edge from-layer="1408" from-port="0" to-layer="1422" to-port="2"/>
+ <edge from-layer="1409" from-port="0" to-layer="1422" to-port="3"/>
+ <edge from-layer="1410" from-port="0" to-layer="1422" to-port="4"/>
+ <edge from-layer="1424" from-port="0" to-layer="1425" to-port="0"/>
+ <edge from-layer="1425" from-port="1" to-layer="1427" to-port="0"/>
+ <edge from-layer="1426" from-port="0" to-layer="1427" to-port="1"/>
+ <edge from-layer="1427" from-port="2" to-layer="1429" to-port="0"/>
+ <edge from-layer="1428" from-port="0" to-layer="1429" to-port="1"/>
+ <edge from-layer="1429" from-port="2" to-layer="1430" to-port="0"/>
+ <edge from-layer="1423" from-port="0" to-layer="1430" to-port="1"/>
+ <edge from-layer="1422" from-port="5" to-layer="1431" to-port="0"/>
+ <edge from-layer="1430" from-port="2" to-layer="1431" to-port="1"/>
+ <edge from-layer="1431" from-port="2" to-layer="1433" to-port="0"/>
+ <edge from-layer="1432" from-port="0" to-layer="1433" to-port="1"/>
+ <edge from-layer="1433" from-port="2" to-layer="1435" to-port="0"/>
+ <edge from-layer="1434" from-port="0" to-layer="1435" to-port="1"/>
+ <edge from-layer="1435" from-port="2" to-layer="1436" to-port="0"/>
+ <edge from-layer="1403" from-port="0" to-layer="1436" to-port="1"/>
+ <edge from-layer="1404" from-port="0" to-layer="1436" to-port="2"/>
+ <edge from-layer="1405" from-port="0" to-layer="1436" to-port="3"/>
+ <edge from-layer="1406" from-port="0" to-layer="1436" to-port="4"/>
+ <edge from-layer="1437" from-port="0" to-layer="1438" to-port="0"/>
+ <edge from-layer="1438" from-port="1" to-layer="1440" to-port="0"/>
+ <edge from-layer="1439" from-port="0" to-layer="1440" to-port="1"/>
+ <edge from-layer="1440" from-port="2" to-layer="1442" to-port="0"/>
+ <edge from-layer="1441" from-port="0" to-layer="1442" to-port="1"/>
+ <edge from-layer="1436" from-port="5" to-layer="1443" to-port="0"/>
+ <edge from-layer="1442" from-port="2" to-layer="1443" to-port="1"/>
+ <edge from-layer="1443" from-port="2" to-layer="1445" to-port="0"/>
+ <edge from-layer="1444" from-port="0" to-layer="1445" to-port="1"/>
+ <edge from-layer="1445" from-port="2" to-layer="1446" to-port="0"/>
+ <edge from-layer="1399" from-port="0" to-layer="1446" to-port="1"/>
+ <edge from-layer="1400" from-port="0" to-layer="1446" to-port="2"/>
+ <edge from-layer="1401" from-port="0" to-layer="1446" to-port="3"/>
+ <edge from-layer="1402" from-port="0" to-layer="1446" to-port="4"/>
+ <edge from-layer="1398" from-port="5" to-layer="1447" to-port="0"/>
+ <edge from-layer="1446" from-port="5" to-layer="1447" to-port="1"/>
+ <edge from-layer="1447" from-port="2" to-layer="1449" to-port="0"/>
+ <edge from-layer="1448" from-port="0" to-layer="1449" to-port="1"/>
+ <edge from-layer="1449" from-port="2" to-layer="1450" to-port="0"/>
+ <edge from-layer="52" from-port="0" to-layer="1450" to-port="1"/>
+ <edge from-layer="53" from-port="0" to-layer="1450" to-port="2"/>
+ <edge from-layer="54" from-port="0" to-layer="1450" to-port="3"/>
+ <edge from-layer="55" from-port="0" to-layer="1450" to-port="4"/>
+ <edge from-layer="1450" from-port="5" to-layer="1451" to-port="0"/>
+ <edge from-layer="1452" from-port="0" to-layer="1453" to-port="0"/>
+ <edge from-layer="1453" from-port="1" to-layer="1455" to-port="0"/>
+ <edge from-layer="1454" from-port="0" to-layer="1455" to-port="1"/>
+ <edge from-layer="1455" from-port="2" to-layer="1457" to-port="0"/>
+ <edge from-layer="1456" from-port="0" to-layer="1457" to-port="1"/>
+ <edge from-layer="1451" from-port="1" to-layer="1458" to-port="0"/>
+ <edge from-layer="1457" from-port="2" to-layer="1458" to-port="1"/>
+ <edge from-layer="1458" from-port="2" to-layer="1460" to-port="0"/>
+ <edge from-layer="1459" from-port="0" to-layer="1460" to-port="1"/>
+ <edge from-layer="1460" from-port="2" to-layer="1461" to-port="0"/>
+ <edge from-layer="48" from-port="0" to-layer="1461" to-port="1"/>
+ <edge from-layer="49" from-port="0" to-layer="1461" to-port="2"/>
+ <edge from-layer="50" from-port="0" to-layer="1461" to-port="3"/>
+ <edge from-layer="51" from-port="0" to-layer="1461" to-port="4"/>
+ <edge from-layer="1474" from-port="0" to-layer="1475" to-port="0"/>
+ <edge from-layer="1475" from-port="1" to-layer="1477" to-port="0"/>
+ <edge from-layer="1476" from-port="0" to-layer="1477" to-port="1"/>
+ <edge from-layer="1477" from-port="2" to-layer="1479" to-port="0"/>
+ <edge from-layer="1478" from-port="0" to-layer="1479" to-port="1"/>
+ <edge from-layer="1450" from-port="5" to-layer="1480" to-port="0"/>
+ <edge from-layer="1479" from-port="2" to-layer="1480" to-port="1"/>
+ <edge from-layer="1480" from-port="2" to-layer="1482" to-port="0"/>
+ <edge from-layer="1481" from-port="0" to-layer="1482" to-port="1"/>
+ <edge from-layer="1482" from-port="2" to-layer="1484" to-port="0"/>
+ <edge from-layer="1483" from-port="0" to-layer="1484" to-port="1"/>
+ <edge from-layer="1484" from-port="2" to-layer="1485" to-port="0"/>
+ <edge from-layer="1470" from-port="0" to-layer="1485" to-port="1"/>
+ <edge from-layer="1471" from-port="0" to-layer="1485" to-port="2"/>
+ <edge from-layer="1472" from-port="0" to-layer="1485" to-port="3"/>
+ <edge from-layer="1473" from-port="0" to-layer="1485" to-port="4"/>
+ <edge from-layer="1487" from-port="0" to-layer="1488" to-port="0"/>
+ <edge from-layer="1488" from-port="1" to-layer="1490" to-port="0"/>
+ <edge from-layer="1489" from-port="0" to-layer="1490" to-port="1"/>
+ <edge from-layer="1490" from-port="2" to-layer="1492" to-port="0"/>
+ <edge from-layer="1491" from-port="0" to-layer="1492" to-port="1"/>
+ <edge from-layer="1492" from-port="2" to-layer="1493" to-port="0"/>
+ <edge from-layer="1486" from-port="0" to-layer="1493" to-port="1"/>
+ <edge from-layer="1485" from-port="5" to-layer="1494" to-port="0"/>
+ <edge from-layer="1493" from-port="2" to-layer="1494" to-port="1"/>
+ <edge from-layer="1494" from-port="2" to-layer="1496" to-port="0"/>
+ <edge from-layer="1495" from-port="0" to-layer="1496" to-port="1"/>
+ <edge from-layer="1496" from-port="2" to-layer="1498" to-port="0"/>
+ <edge from-layer="1497" from-port="0" to-layer="1498" to-port="1"/>
+ <edge from-layer="1498" from-port="2" to-layer="1499" to-port="0"/>
+ <edge from-layer="1466" from-port="0" to-layer="1499" to-port="1"/>
+ <edge from-layer="1467" from-port="0" to-layer="1499" to-port="2"/>
+ <edge from-layer="1468" from-port="0" to-layer="1499" to-port="3"/>
+ <edge from-layer="1469" from-port="0" to-layer="1499" to-port="4"/>
+ <edge from-layer="1500" from-port="0" to-layer="1501" to-port="0"/>
+ <edge from-layer="1501" from-port="1" to-layer="1503" to-port="0"/>
+ <edge from-layer="1502" from-port="0" to-layer="1503" to-port="1"/>
+ <edge from-layer="1503" from-port="2" to-layer="1505" to-port="0"/>
+ <edge from-layer="1504" from-port="0" to-layer="1505" to-port="1"/>
+ <edge from-layer="1499" from-port="5" to-layer="1506" to-port="0"/>
+ <edge from-layer="1505" from-port="2" to-layer="1506" to-port="1"/>
+ <edge from-layer="1506" from-port="2" to-layer="1508" to-port="0"/>
+ <edge from-layer="1507" from-port="0" to-layer="1508" to-port="1"/>
+ <edge from-layer="1508" from-port="2" to-layer="1509" to-port="0"/>
+ <edge from-layer="1462" from-port="0" to-layer="1509" to-port="1"/>
+ <edge from-layer="1463" from-port="0" to-layer="1509" to-port="2"/>
+ <edge from-layer="1464" from-port="0" to-layer="1509" to-port="3"/>
+ <edge from-layer="1465" from-port="0" to-layer="1509" to-port="4"/>
+ <edge from-layer="1461" from-port="5" to-layer="1510" to-port="0"/>
+ <edge from-layer="1509" from-port="5" to-layer="1510" to-port="1"/>
+ <edge from-layer="1510" from-port="2" to-layer="1512" to-port="0"/>
+ <edge from-layer="1511" from-port="0" to-layer="1512" to-port="1"/>
+ <edge from-layer="1512" from-port="2" to-layer="1513" to-port="0"/>
+ <edge from-layer="44" from-port="0" to-layer="1513" to-port="1"/>
+ <edge from-layer="45" from-port="0" to-layer="1513" to-port="2"/>
+ <edge from-layer="46" from-port="0" to-layer="1513" to-port="3"/>
+ <edge from-layer="47" from-port="0" to-layer="1513" to-port="4"/>
+ <edge from-layer="1526" from-port="0" to-layer="1527" to-port="0"/>
+ <edge from-layer="1527" from-port="1" to-layer="1529" to-port="0"/>
+ <edge from-layer="1528" from-port="0" to-layer="1529" to-port="1"/>
+ <edge from-layer="1529" from-port="2" to-layer="1531" to-port="0"/>
+ <edge from-layer="1530" from-port="0" to-layer="1531" to-port="1"/>
+ <edge from-layer="1513" from-port="5" to-layer="1532" to-port="0"/>
+ <edge from-layer="1531" from-port="2" to-layer="1532" to-port="1"/>
+ <edge from-layer="1532" from-port="2" to-layer="1534" to-port="0"/>
+ <edge from-layer="1533" from-port="0" to-layer="1534" to-port="1"/>
+ <edge from-layer="1534" from-port="2" to-layer="1536" to-port="0"/>
+ <edge from-layer="1535" from-port="0" to-layer="1536" to-port="1"/>
+ <edge from-layer="1536" from-port="2" to-layer="1537" to-port="0"/>
+ <edge from-layer="1522" from-port="0" to-layer="1537" to-port="1"/>
+ <edge from-layer="1523" from-port="0" to-layer="1537" to-port="2"/>
+ <edge from-layer="1524" from-port="0" to-layer="1537" to-port="3"/>
+ <edge from-layer="1525" from-port="0" to-layer="1537" to-port="4"/>
+ <edge from-layer="1539" from-port="0" to-layer="1540" to-port="0"/>
+ <edge from-layer="1540" from-port="1" to-layer="1542" to-port="0"/>
+ <edge from-layer="1541" from-port="0" to-layer="1542" to-port="1"/>
+ <edge from-layer="1542" from-port="2" to-layer="1544" to-port="0"/>
+ <edge from-layer="1543" from-port="0" to-layer="1544" to-port="1"/>
+ <edge from-layer="1544" from-port="2" to-layer="1545" to-port="0"/>
+ <edge from-layer="1538" from-port="0" to-layer="1545" to-port="1"/>
+ <edge from-layer="1537" from-port="5" to-layer="1546" to-port="0"/>
+ <edge from-layer="1545" from-port="2" to-layer="1546" to-port="1"/>
+ <edge from-layer="1546" from-port="2" to-layer="1548" to-port="0"/>
+ <edge from-layer="1547" from-port="0" to-layer="1548" to-port="1"/>
+ <edge from-layer="1548" from-port="2" to-layer="1550" to-port="0"/>
+ <edge from-layer="1549" from-port="0" to-layer="1550" to-port="1"/>
+ <edge from-layer="1550" from-port="2" to-layer="1551" to-port="0"/>
+ <edge from-layer="1518" from-port="0" to-layer="1551" to-port="1"/>
+ <edge from-layer="1519" from-port="0" to-layer="1551" to-port="2"/>
+ <edge from-layer="1520" from-port="0" to-layer="1551" to-port="3"/>
+ <edge from-layer="1521" from-port="0" to-layer="1551" to-port="4"/>
+ <edge from-layer="1552" from-port="0" to-layer="1553" to-port="0"/>
+ <edge from-layer="1553" from-port="1" to-layer="1555" to-port="0"/>
+ <edge from-layer="1554" from-port="0" to-layer="1555" to-port="1"/>
+ <edge from-layer="1555" from-port="2" to-layer="1557" to-port="0"/>
+ <edge from-layer="1556" from-port="0" to-layer="1557" to-port="1"/>
+ <edge from-layer="1551" from-port="5" to-layer="1558" to-port="0"/>
+ <edge from-layer="1557" from-port="2" to-layer="1558" to-port="1"/>
+ <edge from-layer="1558" from-port="2" to-layer="1560" to-port="0"/>
+ <edge from-layer="1559" from-port="0" to-layer="1560" to-port="1"/>
+ <edge from-layer="1560" from-port="2" to-layer="1561" to-port="0"/>
+ <edge from-layer="1514" from-port="0" to-layer="1561" to-port="1"/>
+ <edge from-layer="1515" from-port="0" to-layer="1561" to-port="2"/>
+ <edge from-layer="1516" from-port="0" to-layer="1561" to-port="3"/>
+ <edge from-layer="1517" from-port="0" to-layer="1561" to-port="4"/>
+ <edge from-layer="1513" from-port="5" to-layer="1562" to-port="0"/>
+ <edge from-layer="1561" from-port="5" to-layer="1562" to-port="1"/>
+ <edge from-layer="1562" from-port="2" to-layer="1564" to-port="0"/>
+ <edge from-layer="1563" from-port="0" to-layer="1564" to-port="1"/>
+ <edge from-layer="1564" from-port="2" to-layer="1565" to-port="0"/>
+ <edge from-layer="40" from-port="0" to-layer="1565" to-port="1"/>
+ <edge from-layer="41" from-port="0" to-layer="1565" to-port="2"/>
+ <edge from-layer="42" from-port="0" to-layer="1565" to-port="3"/>
+ <edge from-layer="43" from-port="0" to-layer="1565" to-port="4"/>
+ <edge from-layer="1578" from-port="0" to-layer="1579" to-port="0"/>
+ <edge from-layer="1579" from-port="1" to-layer="1581" to-port="0"/>
+ <edge from-layer="1580" from-port="0" to-layer="1581" to-port="1"/>
+ <edge from-layer="1581" from-port="2" to-layer="1583" to-port="0"/>
+ <edge from-layer="1582" from-port="0" to-layer="1583" to-port="1"/>
+ <edge from-layer="1565" from-port="5" to-layer="1584" to-port="0"/>
+ <edge from-layer="1583" from-port="2" to-layer="1584" to-port="1"/>
+ <edge from-layer="1584" from-port="2" to-layer="1586" to-port="0"/>
+ <edge from-layer="1585" from-port="0" to-layer="1586" to-port="1"/>
+ <edge from-layer="1586" from-port="2" to-layer="1588" to-port="0"/>
+ <edge from-layer="1587" from-port="0" to-layer="1588" to-port="1"/>
+ <edge from-layer="1588" from-port="2" to-layer="1589" to-port="0"/>
+ <edge from-layer="1574" from-port="0" to-layer="1589" to-port="1"/>
+ <edge from-layer="1575" from-port="0" to-layer="1589" to-port="2"/>
+ <edge from-layer="1576" from-port="0" to-layer="1589" to-port="3"/>
+ <edge from-layer="1577" from-port="0" to-layer="1589" to-port="4"/>
+ <edge from-layer="1591" from-port="0" to-layer="1592" to-port="0"/>
+ <edge from-layer="1592" from-port="1" to-layer="1594" to-port="0"/>
+ <edge from-layer="1593" from-port="0" to-layer="1594" to-port="1"/>
+ <edge from-layer="1594" from-port="2" to-layer="1596" to-port="0"/>
+ <edge from-layer="1595" from-port="0" to-layer="1596" to-port="1"/>
+ <edge from-layer="1596" from-port="2" to-layer="1597" to-port="0"/>
+ <edge from-layer="1590" from-port="0" to-layer="1597" to-port="1"/>
+ <edge from-layer="1589" from-port="5" to-layer="1598" to-port="0"/>
+ <edge from-layer="1597" from-port="2" to-layer="1598" to-port="1"/>
+ <edge from-layer="1598" from-port="2" to-layer="1600" to-port="0"/>
+ <edge from-layer="1599" from-port="0" to-layer="1600" to-port="1"/>
+ <edge from-layer="1600" from-port="2" to-layer="1602" to-port="0"/>
+ <edge from-layer="1601" from-port="0" to-layer="1602" to-port="1"/>
+ <edge from-layer="1602" from-port="2" to-layer="1603" to-port="0"/>
+ <edge from-layer="1570" from-port="0" to-layer="1603" to-port="1"/>
+ <edge from-layer="1571" from-port="0" to-layer="1603" to-port="2"/>
+ <edge from-layer="1572" from-port="0" to-layer="1603" to-port="3"/>
+ <edge from-layer="1573" from-port="0" to-layer="1603" to-port="4"/>
+ <edge from-layer="1604" from-port="0" to-layer="1605" to-port="0"/>
+ <edge from-layer="1605" from-port="1" to-layer="1607" to-port="0"/>
+ <edge from-layer="1606" from-port="0" to-layer="1607" to-port="1"/>
+ <edge from-layer="1607" from-port="2" to-layer="1609" to-port="0"/>
+ <edge from-layer="1608" from-port="0" to-layer="1609" to-port="1"/>
+ <edge from-layer="1603" from-port="5" to-layer="1610" to-port="0"/>
+ <edge from-layer="1609" from-port="2" to-layer="1610" to-port="1"/>
+ <edge from-layer="1610" from-port="2" to-layer="1612" to-port="0"/>
+ <edge from-layer="1611" from-port="0" to-layer="1612" to-port="1"/>
+ <edge from-layer="1612" from-port="2" to-layer="1613" to-port="0"/>
+ <edge from-layer="1566" from-port="0" to-layer="1613" to-port="1"/>
+ <edge from-layer="1567" from-port="0" to-layer="1613" to-port="2"/>
+ <edge from-layer="1568" from-port="0" to-layer="1613" to-port="3"/>
+ <edge from-layer="1569" from-port="0" to-layer="1613" to-port="4"/>
+ <edge from-layer="1565" from-port="5" to-layer="1614" to-port="0"/>
+ <edge from-layer="1613" from-port="5" to-layer="1614" to-port="1"/>
+ <edge from-layer="1614" from-port="2" to-layer="1616" to-port="0"/>
+ <edge from-layer="1615" from-port="0" to-layer="1616" to-port="1"/>
+ <edge from-layer="1616" from-port="2" to-layer="1617" to-port="0"/>
+ <edge from-layer="36" from-port="0" to-layer="1617" to-port="1"/>
+ <edge from-layer="37" from-port="0" to-layer="1617" to-port="2"/>
+ <edge from-layer="38" from-port="0" to-layer="1617" to-port="3"/>
+ <edge from-layer="39" from-port="0" to-layer="1617" to-port="4"/>
+ <edge from-layer="1630" from-port="0" to-layer="1631" to-port="0"/>
+ <edge from-layer="1631" from-port="1" to-layer="1633" to-port="0"/>
+ <edge from-layer="1632" from-port="0" to-layer="1633" to-port="1"/>
+ <edge from-layer="1633" from-port="2" to-layer="1635" to-port="0"/>
+ <edge from-layer="1634" from-port="0" to-layer="1635" to-port="1"/>
+ <edge from-layer="1617" from-port="5" to-layer="1636" to-port="0"/>
+ <edge from-layer="1635" from-port="2" to-layer="1636" to-port="1"/>
+ <edge from-layer="1636" from-port="2" to-layer="1638" to-port="0"/>
+ <edge from-layer="1637" from-port="0" to-layer="1638" to-port="1"/>
+ <edge from-layer="1638" from-port="2" to-layer="1640" to-port="0"/>
+ <edge from-layer="1639" from-port="0" to-layer="1640" to-port="1"/>
+ <edge from-layer="1640" from-port="2" to-layer="1641" to-port="0"/>
+ <edge from-layer="1626" from-port="0" to-layer="1641" to-port="1"/>
+ <edge from-layer="1627" from-port="0" to-layer="1641" to-port="2"/>
+ <edge from-layer="1628" from-port="0" to-layer="1641" to-port="3"/>
+ <edge from-layer="1629" from-port="0" to-layer="1641" to-port="4"/>
+ <edge from-layer="1643" from-port="0" to-layer="1644" to-port="0"/>
+ <edge from-layer="1644" from-port="1" to-layer="1646" to-port="0"/>
+ <edge from-layer="1645" from-port="0" to-layer="1646" to-port="1"/>
+ <edge from-layer="1646" from-port="2" to-layer="1648" to-port="0"/>
+ <edge from-layer="1647" from-port="0" to-layer="1648" to-port="1"/>
+ <edge from-layer="1648" from-port="2" to-layer="1649" to-port="0"/>
+ <edge from-layer="1642" from-port="0" to-layer="1649" to-port="1"/>
+ <edge from-layer="1641" from-port="5" to-layer="1650" to-port="0"/>
+ <edge from-layer="1649" from-port="2" to-layer="1650" to-port="1"/>
+ <edge from-layer="1650" from-port="2" to-layer="1652" to-port="0"/>
+ <edge from-layer="1651" from-port="0" to-layer="1652" to-port="1"/>
+ <edge from-layer="1652" from-port="2" to-layer="1654" to-port="0"/>
+ <edge from-layer="1653" from-port="0" to-layer="1654" to-port="1"/>
+ <edge from-layer="1654" from-port="2" to-layer="1655" to-port="0"/>
+ <edge from-layer="1622" from-port="0" to-layer="1655" to-port="1"/>
+ <edge from-layer="1623" from-port="0" to-layer="1655" to-port="2"/>
+ <edge from-layer="1624" from-port="0" to-layer="1655" to-port="3"/>
+ <edge from-layer="1625" from-port="0" to-layer="1655" to-port="4"/>
+ <edge from-layer="1656" from-port="0" to-layer="1657" to-port="0"/>
+ <edge from-layer="1657" from-port="1" to-layer="1659" to-port="0"/>
+ <edge from-layer="1658" from-port="0" to-layer="1659" to-port="1"/>
+ <edge from-layer="1659" from-port="2" to-layer="1661" to-port="0"/>
+ <edge from-layer="1660" from-port="0" to-layer="1661" to-port="1"/>
+ <edge from-layer="1655" from-port="5" to-layer="1662" to-port="0"/>
+ <edge from-layer="1661" from-port="2" to-layer="1662" to-port="1"/>
+ <edge from-layer="1662" from-port="2" to-layer="1664" to-port="0"/>
+ <edge from-layer="1663" from-port="0" to-layer="1664" to-port="1"/>
+ <edge from-layer="1664" from-port="2" to-layer="1665" to-port="0"/>
+ <edge from-layer="1618" from-port="0" to-layer="1665" to-port="1"/>
+ <edge from-layer="1619" from-port="0" to-layer="1665" to-port="2"/>
+ <edge from-layer="1620" from-port="0" to-layer="1665" to-port="3"/>
+ <edge from-layer="1621" from-port="0" to-layer="1665" to-port="4"/>
+ <edge from-layer="1617" from-port="5" to-layer="1666" to-port="0"/>
+ <edge from-layer="1665" from-port="5" to-layer="1666" to-port="1"/>
+ <edge from-layer="1666" from-port="2" to-layer="1668" to-port="0"/>
+ <edge from-layer="1667" from-port="0" to-layer="1668" to-port="1"/>
+ <edge from-layer="1668" from-port="2" to-layer="1669" to-port="0"/>
+ <edge from-layer="32" from-port="0" to-layer="1669" to-port="1"/>
+ <edge from-layer="33" from-port="0" to-layer="1669" to-port="2"/>
+ <edge from-layer="34" from-port="0" to-layer="1669" to-port="3"/>
+ <edge from-layer="35" from-port="0" to-layer="1669" to-port="4"/>
+ <edge from-layer="1682" from-port="0" to-layer="1683" to-port="0"/>
+ <edge from-layer="1683" from-port="1" to-layer="1685" to-port="0"/>
+ <edge from-layer="1684" from-port="0" to-layer="1685" to-port="1"/>
+ <edge from-layer="1685" from-port="2" to-layer="1687" to-port="0"/>
+ <edge from-layer="1686" from-port="0" to-layer="1687" to-port="1"/>
+ <edge from-layer="1669" from-port="5" to-layer="1688" to-port="0"/>
+ <edge from-layer="1687" from-port="2" to-layer="1688" to-port="1"/>
+ <edge from-layer="1688" from-port="2" to-layer="1690" to-port="0"/>
+ <edge from-layer="1689" from-port="0" to-layer="1690" to-port="1"/>
+ <edge from-layer="1690" from-port="2" to-layer="1692" to-port="0"/>
+ <edge from-layer="1691" from-port="0" to-layer="1692" to-port="1"/>
+ <edge from-layer="1692" from-port="2" to-layer="1693" to-port="0"/>
+ <edge from-layer="1678" from-port="0" to-layer="1693" to-port="1"/>
+ <edge from-layer="1679" from-port="0" to-layer="1693" to-port="2"/>
+ <edge from-layer="1680" from-port="0" to-layer="1693" to-port="3"/>
+ <edge from-layer="1681" from-port="0" to-layer="1693" to-port="4"/>
+ <edge from-layer="1695" from-port="0" to-layer="1696" to-port="0"/>
+ <edge from-layer="1696" from-port="1" to-layer="1698" to-port="0"/>
+ <edge from-layer="1697" from-port="0" to-layer="1698" to-port="1"/>
+ <edge from-layer="1698" from-port="2" to-layer="1700" to-port="0"/>
+ <edge from-layer="1699" from-port="0" to-layer="1700" to-port="1"/>
+ <edge from-layer="1700" from-port="2" to-layer="1701" to-port="0"/>
+ <edge from-layer="1694" from-port="0" to-layer="1701" to-port="1"/>
+ <edge from-layer="1693" from-port="5" to-layer="1702" to-port="0"/>
+ <edge from-layer="1701" from-port="2" to-layer="1702" to-port="1"/>
+ <edge from-layer="1702" from-port="2" to-layer="1704" to-port="0"/>
+ <edge from-layer="1703" from-port="0" to-layer="1704" to-port="1"/>
+ <edge from-layer="1704" from-port="2" to-layer="1706" to-port="0"/>
+ <edge from-layer="1705" from-port="0" to-layer="1706" to-port="1"/>
+ <edge from-layer="1706" from-port="2" to-layer="1707" to-port="0"/>
+ <edge from-layer="1674" from-port="0" to-layer="1707" to-port="1"/>
+ <edge from-layer="1675" from-port="0" to-layer="1707" to-port="2"/>
+ <edge from-layer="1676" from-port="0" to-layer="1707" to-port="3"/>
+ <edge from-layer="1677" from-port="0" to-layer="1707" to-port="4"/>
+ <edge from-layer="1708" from-port="0" to-layer="1709" to-port="0"/>
+ <edge from-layer="1709" from-port="1" to-layer="1711" to-port="0"/>
+ <edge from-layer="1710" from-port="0" to-layer="1711" to-port="1"/>
+ <edge from-layer="1711" from-port="2" to-layer="1713" to-port="0"/>
+ <edge from-layer="1712" from-port="0" to-layer="1713" to-port="1"/>
+ <edge from-layer="1707" from-port="5" to-layer="1714" to-port="0"/>
+ <edge from-layer="1713" from-port="2" to-layer="1714" to-port="1"/>
+ <edge from-layer="1714" from-port="2" to-layer="1716" to-port="0"/>
+ <edge from-layer="1715" from-port="0" to-layer="1716" to-port="1"/>
+ <edge from-layer="1716" from-port="2" to-layer="1717" to-port="0"/>
+ <edge from-layer="1670" from-port="0" to-layer="1717" to-port="1"/>
+ <edge from-layer="1671" from-port="0" to-layer="1717" to-port="2"/>
+ <edge from-layer="1672" from-port="0" to-layer="1717" to-port="3"/>
+ <edge from-layer="1673" from-port="0" to-layer="1717" to-port="4"/>
+ <edge from-layer="1669" from-port="5" to-layer="1718" to-port="0"/>
+ <edge from-layer="1717" from-port="5" to-layer="1718" to-port="1"/>
+ <edge from-layer="1718" from-port="2" to-layer="1720" to-port="0"/>
+ <edge from-layer="1719" from-port="0" to-layer="1720" to-port="1"/>
+ <edge from-layer="1720" from-port="2" to-layer="1721" to-port="0"/>
+ <edge from-layer="28" from-port="0" to-layer="1721" to-port="1"/>
+ <edge from-layer="29" from-port="0" to-layer="1721" to-port="2"/>
+ <edge from-layer="30" from-port="0" to-layer="1721" to-port="3"/>
+ <edge from-layer="31" from-port="0" to-layer="1721" to-port="4"/>
+ <edge from-layer="1734" from-port="0" to-layer="1735" to-port="0"/>
+ <edge from-layer="1735" from-port="1" to-layer="1737" to-port="0"/>
+ <edge from-layer="1736" from-port="0" to-layer="1737" to-port="1"/>
+ <edge from-layer="1737" from-port="2" to-layer="1739" to-port="0"/>
+ <edge from-layer="1738" from-port="0" to-layer="1739" to-port="1"/>
+ <edge from-layer="1721" from-port="5" to-layer="1740" to-port="0"/>
+ <edge from-layer="1739" from-port="2" to-layer="1740" to-port="1"/>
+ <edge from-layer="1740" from-port="2" to-layer="1742" to-port="0"/>
+ <edge from-layer="1741" from-port="0" to-layer="1742" to-port="1"/>
+ <edge from-layer="1742" from-port="2" to-layer="1744" to-port="0"/>
+ <edge from-layer="1743" from-port="0" to-layer="1744" to-port="1"/>
+ <edge from-layer="1744" from-port="2" to-layer="1745" to-port="0"/>
+ <edge from-layer="1730" from-port="0" to-layer="1745" to-port="1"/>
+ <edge from-layer="1731" from-port="0" to-layer="1745" to-port="2"/>
+ <edge from-layer="1732" from-port="0" to-layer="1745" to-port="3"/>
+ <edge from-layer="1733" from-port="0" to-layer="1745" to-port="4"/>
+ <edge from-layer="1747" from-port="0" to-layer="1748" to-port="0"/>
+ <edge from-layer="1748" from-port="1" to-layer="1750" to-port="0"/>
+ <edge from-layer="1749" from-port="0" to-layer="1750" to-port="1"/>
+ <edge from-layer="1750" from-port="2" to-layer="1752" to-port="0"/>
+ <edge from-layer="1751" from-port="0" to-layer="1752" to-port="1"/>
+ <edge from-layer="1752" from-port="2" to-layer="1753" to-port="0"/>
+ <edge from-layer="1746" from-port="0" to-layer="1753" to-port="1"/>
+ <edge from-layer="1745" from-port="5" to-layer="1754" to-port="0"/>
+ <edge from-layer="1753" from-port="2" to-layer="1754" to-port="1"/>
+ <edge from-layer="1754" from-port="2" to-layer="1756" to-port="0"/>
+ <edge from-layer="1755" from-port="0" to-layer="1756" to-port="1"/>
+ <edge from-layer="1756" from-port="2" to-layer="1758" to-port="0"/>
+ <edge from-layer="1757" from-port="0" to-layer="1758" to-port="1"/>
+ <edge from-layer="1758" from-port="2" to-layer="1759" to-port="0"/>
+ <edge from-layer="1726" from-port="0" to-layer="1759" to-port="1"/>
+ <edge from-layer="1727" from-port="0" to-layer="1759" to-port="2"/>
+ <edge from-layer="1728" from-port="0" to-layer="1759" to-port="3"/>
+ <edge from-layer="1729" from-port="0" to-layer="1759" to-port="4"/>
+ <edge from-layer="1760" from-port="0" to-layer="1761" to-port="0"/>
+ <edge from-layer="1761" from-port="1" to-layer="1763" to-port="0"/>
+ <edge from-layer="1762" from-port="0" to-layer="1763" to-port="1"/>
+ <edge from-layer="1763" from-port="2" to-layer="1765" to-port="0"/>
+ <edge from-layer="1764" from-port="0" to-layer="1765" to-port="1"/>
+ <edge from-layer="1759" from-port="5" to-layer="1766" to-port="0"/>
+ <edge from-layer="1765" from-port="2" to-layer="1766" to-port="1"/>
+ <edge from-layer="1766" from-port="2" to-layer="1768" to-port="0"/>
+ <edge from-layer="1767" from-port="0" to-layer="1768" to-port="1"/>
+ <edge from-layer="1768" from-port="2" to-layer="1769" to-port="0"/>
+ <edge from-layer="1722" from-port="0" to-layer="1769" to-port="1"/>
+ <edge from-layer="1723" from-port="0" to-layer="1769" to-port="2"/>
+ <edge from-layer="1724" from-port="0" to-layer="1769" to-port="3"/>
+ <edge from-layer="1725" from-port="0" to-layer="1769" to-port="4"/>
+ <edge from-layer="1721" from-port="5" to-layer="1770" to-port="0"/>
+ <edge from-layer="1769" from-port="5" to-layer="1770" to-port="1"/>
+ <edge from-layer="1770" from-port="2" to-layer="1772" to-port="0"/>
+ <edge from-layer="1771" from-port="0" to-layer="1772" to-port="1"/>
+ <edge from-layer="1772" from-port="2" to-layer="1773" to-port="0"/>
+ <edge from-layer="24" from-port="0" to-layer="1773" to-port="1"/>
+ <edge from-layer="25" from-port="0" to-layer="1773" to-port="2"/>
+ <edge from-layer="26" from-port="0" to-layer="1773" to-port="3"/>
+ <edge from-layer="27" from-port="0" to-layer="1773" to-port="4"/>
+ <edge from-layer="1786" from-port="0" to-layer="1787" to-port="0"/>
+ <edge from-layer="1787" from-port="1" to-layer="1789" to-port="0"/>
+ <edge from-layer="1788" from-port="0" to-layer="1789" to-port="1"/>
+ <edge from-layer="1789" from-port="2" to-layer="1791" to-port="0"/>
+ <edge from-layer="1790" from-port="0" to-layer="1791" to-port="1"/>
+ <edge from-layer="1773" from-port="5" to-layer="1792" to-port="0"/>
+ <edge from-layer="1791" from-port="2" to-layer="1792" to-port="1"/>
+ <edge from-layer="1792" from-port="2" to-layer="1794" to-port="0"/>
+ <edge from-layer="1793" from-port="0" to-layer="1794" to-port="1"/>
+ <edge from-layer="1794" from-port="2" to-layer="1796" to-port="0"/>
+ <edge from-layer="1795" from-port="0" to-layer="1796" to-port="1"/>
+ <edge from-layer="1796" from-port="2" to-layer="1797" to-port="0"/>
+ <edge from-layer="1782" from-port="0" to-layer="1797" to-port="1"/>
+ <edge from-layer="1783" from-port="0" to-layer="1797" to-port="2"/>
+ <edge from-layer="1784" from-port="0" to-layer="1797" to-port="3"/>
+ <edge from-layer="1785" from-port="0" to-layer="1797" to-port="4"/>
+ <edge from-layer="1799" from-port="0" to-layer="1800" to-port="0"/>
+ <edge from-layer="1800" from-port="1" to-layer="1802" to-port="0"/>
+ <edge from-layer="1801" from-port="0" to-layer="1802" to-port="1"/>
+ <edge from-layer="1802" from-port="2" to-layer="1804" to-port="0"/>
+ <edge from-layer="1803" from-port="0" to-layer="1804" to-port="1"/>
+ <edge from-layer="1804" from-port="2" to-layer="1805" to-port="0"/>
+ <edge from-layer="1798" from-port="0" to-layer="1805" to-port="1"/>
+ <edge from-layer="1797" from-port="5" to-layer="1806" to-port="0"/>
+ <edge from-layer="1805" from-port="2" to-layer="1806" to-port="1"/>
+ <edge from-layer="1806" from-port="2" to-layer="1808" to-port="0"/>
+ <edge from-layer="1807" from-port="0" to-layer="1808" to-port="1"/>
+ <edge from-layer="1808" from-port="2" to-layer="1810" to-port="0"/>
+ <edge from-layer="1809" from-port="0" to-layer="1810" to-port="1"/>
+ <edge from-layer="1810" from-port="2" to-layer="1811" to-port="0"/>
+ <edge from-layer="1778" from-port="0" to-layer="1811" to-port="1"/>
+ <edge from-layer="1779" from-port="0" to-layer="1811" to-port="2"/>
+ <edge from-layer="1780" from-port="0" to-layer="1811" to-port="3"/>
+ <edge from-layer="1781" from-port="0" to-layer="1811" to-port="4"/>
+ <edge from-layer="1812" from-port="0" to-layer="1813" to-port="0"/>
+ <edge from-layer="1813" from-port="1" to-layer="1815" to-port="0"/>
+ <edge from-layer="1814" from-port="0" to-layer="1815" to-port="1"/>
+ <edge from-layer="1815" from-port="2" to-layer="1817" to-port="0"/>
+ <edge from-layer="1816" from-port="0" to-layer="1817" to-port="1"/>
+ <edge from-layer="1811" from-port="5" to-layer="1818" to-port="0"/>
+ <edge from-layer="1817" from-port="2" to-layer="1818" to-port="1"/>
+ <edge from-layer="1818" from-port="2" to-layer="1820" to-port="0"/>
+ <edge from-layer="1819" from-port="0" to-layer="1820" to-port="1"/>
+ <edge from-layer="1820" from-port="2" to-layer="1821" to-port="0"/>
+ <edge from-layer="1774" from-port="0" to-layer="1821" to-port="1"/>
+ <edge from-layer="1775" from-port="0" to-layer="1821" to-port="2"/>
+ <edge from-layer="1776" from-port="0" to-layer="1821" to-port="3"/>
+ <edge from-layer="1777" from-port="0" to-layer="1821" to-port="4"/>
+ <edge from-layer="1773" from-port="5" to-layer="1822" to-port="0"/>
+ <edge from-layer="1821" from-port="5" to-layer="1822" to-port="1"/>
+ <edge from-layer="1822" from-port="2" to-layer="1824" to-port="0"/>
+ <edge from-layer="1823" from-port="0" to-layer="1824" to-port="1"/>
+ <edge from-layer="1824" from-port="2" to-layer="1825" to-port="0"/>
+ <edge from-layer="20" from-port="0" to-layer="1825" to-port="1"/>
+ <edge from-layer="21" from-port="0" to-layer="1825" to-port="2"/>
+ <edge from-layer="22" from-port="0" to-layer="1825" to-port="3"/>
+ <edge from-layer="23" from-port="0" to-layer="1825" to-port="4"/>
+ <edge from-layer="1838" from-port="0" to-layer="1839" to-port="0"/>
+ <edge from-layer="1839" from-port="1" to-layer="1841" to-port="0"/>
+ <edge from-layer="1840" from-port="0" to-layer="1841" to-port="1"/>
+ <edge from-layer="1841" from-port="2" to-layer="1843" to-port="0"/>
+ <edge from-layer="1842" from-port="0" to-layer="1843" to-port="1"/>
+ <edge from-layer="1825" from-port="5" to-layer="1844" to-port="0"/>
+ <edge from-layer="1843" from-port="2" to-layer="1844" to-port="1"/>
+ <edge from-layer="1844" from-port="2" to-layer="1846" to-port="0"/>
+ <edge from-layer="1845" from-port="0" to-layer="1846" to-port="1"/>
+ <edge from-layer="1846" from-port="2" to-layer="1848" to-port="0"/>
+ <edge from-layer="1847" from-port="0" to-layer="1848" to-port="1"/>
+ <edge from-layer="1848" from-port="2" to-layer="1849" to-port="0"/>
+ <edge from-layer="1834" from-port="0" to-layer="1849" to-port="1"/>
+ <edge from-layer="1835" from-port="0" to-layer="1849" to-port="2"/>
+ <edge from-layer="1836" from-port="0" to-layer="1849" to-port="3"/>
+ <edge from-layer="1837" from-port="0" to-layer="1849" to-port="4"/>
+ <edge from-layer="1851" from-port="0" to-layer="1852" to-port="0"/>
+ <edge from-layer="1852" from-port="1" to-layer="1854" to-port="0"/>
+ <edge from-layer="1853" from-port="0" to-layer="1854" to-port="1"/>
+ <edge from-layer="1854" from-port="2" to-layer="1856" to-port="0"/>
+ <edge from-layer="1855" from-port="0" to-layer="1856" to-port="1"/>
+ <edge from-layer="1856" from-port="2" to-layer="1857" to-port="0"/>
+ <edge from-layer="1850" from-port="0" to-layer="1857" to-port="1"/>
+ <edge from-layer="1849" from-port="5" to-layer="1858" to-port="0"/>
+ <edge from-layer="1857" from-port="2" to-layer="1858" to-port="1"/>
+ <edge from-layer="1858" from-port="2" to-layer="1860" to-port="0"/>
+ <edge from-layer="1859" from-port="0" to-layer="1860" to-port="1"/>
+ <edge from-layer="1860" from-port="2" to-layer="1862" to-port="0"/>
+ <edge from-layer="1861" from-port="0" to-layer="1862" to-port="1"/>
+ <edge from-layer="1862" from-port="2" to-layer="1863" to-port="0"/>
+ <edge from-layer="1830" from-port="0" to-layer="1863" to-port="1"/>
+ <edge from-layer="1831" from-port="0" to-layer="1863" to-port="2"/>
+ <edge from-layer="1832" from-port="0" to-layer="1863" to-port="3"/>
+ <edge from-layer="1833" from-port="0" to-layer="1863" to-port="4"/>
+ <edge from-layer="1864" from-port="0" to-layer="1865" to-port="0"/>
+ <edge from-layer="1865" from-port="1" to-layer="1867" to-port="0"/>
+ <edge from-layer="1866" from-port="0" to-layer="1867" to-port="1"/>
+ <edge from-layer="1867" from-port="2" to-layer="1869" to-port="0"/>
+ <edge from-layer="1868" from-port="0" to-layer="1869" to-port="1"/>
+ <edge from-layer="1863" from-port="5" to-layer="1870" to-port="0"/>
+ <edge from-layer="1869" from-port="2" to-layer="1870" to-port="1"/>
+ <edge from-layer="1870" from-port="2" to-layer="1872" to-port="0"/>
+ <edge from-layer="1871" from-port="0" to-layer="1872" to-port="1"/>
+ <edge from-layer="1872" from-port="2" to-layer="1873" to-port="0"/>
+ <edge from-layer="1826" from-port="0" to-layer="1873" to-port="1"/>
+ <edge from-layer="1827" from-port="0" to-layer="1873" to-port="2"/>
+ <edge from-layer="1828" from-port="0" to-layer="1873" to-port="3"/>
+ <edge from-layer="1829" from-port="0" to-layer="1873" to-port="4"/>
+ <edge from-layer="1825" from-port="5" to-layer="1874" to-port="0"/>
+ <edge from-layer="1873" from-port="5" to-layer="1874" to-port="1"/>
+ <edge from-layer="1874" from-port="2" to-layer="1876" to-port="0"/>
+ <edge from-layer="1875" from-port="0" to-layer="1876" to-port="1"/>
+ <edge from-layer="1876" from-port="2" to-layer="1877" to-port="0"/>
+ <edge from-layer="16" from-port="0" to-layer="1877" to-port="1"/>
+ <edge from-layer="17" from-port="0" to-layer="1877" to-port="2"/>
+ <edge from-layer="18" from-port="0" to-layer="1877" to-port="3"/>
+ <edge from-layer="19" from-port="0" to-layer="1877" to-port="4"/>
+ <edge from-layer="1890" from-port="0" to-layer="1891" to-port="0"/>
+ <edge from-layer="1891" from-port="1" to-layer="1893" to-port="0"/>
+ <edge from-layer="1892" from-port="0" to-layer="1893" to-port="1"/>
+ <edge from-layer="1893" from-port="2" to-layer="1895" to-port="0"/>
+ <edge from-layer="1894" from-port="0" to-layer="1895" to-port="1"/>
+ <edge from-layer="1877" from-port="5" to-layer="1896" to-port="0"/>
+ <edge from-layer="1895" from-port="2" to-layer="1896" to-port="1"/>
+ <edge from-layer="1896" from-port="2" to-layer="1898" to-port="0"/>
+ <edge from-layer="1897" from-port="0" to-layer="1898" to-port="1"/>
+ <edge from-layer="1898" from-port="2" to-layer="1900" to-port="0"/>
+ <edge from-layer="1899" from-port="0" to-layer="1900" to-port="1"/>
+ <edge from-layer="1900" from-port="2" to-layer="1901" to-port="0"/>
+ <edge from-layer="1886" from-port="0" to-layer="1901" to-port="1"/>
+ <edge from-layer="1887" from-port="0" to-layer="1901" to-port="2"/>
+ <edge from-layer="1888" from-port="0" to-layer="1901" to-port="3"/>
+ <edge from-layer="1889" from-port="0" to-layer="1901" to-port="4"/>
+ <edge from-layer="1903" from-port="0" to-layer="1904" to-port="0"/>
+ <edge from-layer="1904" from-port="1" to-layer="1906" to-port="0"/>
+ <edge from-layer="1905" from-port="0" to-layer="1906" to-port="1"/>
+ <edge from-layer="1906" from-port="2" to-layer="1908" to-port="0"/>
+ <edge from-layer="1907" from-port="0" to-layer="1908" to-port="1"/>
+ <edge from-layer="1908" from-port="2" to-layer="1909" to-port="0"/>
+ <edge from-layer="1902" from-port="0" to-layer="1909" to-port="1"/>
+ <edge from-layer="1901" from-port="5" to-layer="1910" to-port="0"/>
+ <edge from-layer="1909" from-port="2" to-layer="1910" to-port="1"/>
+ <edge from-layer="1910" from-port="2" to-layer="1912" to-port="0"/>
+ <edge from-layer="1911" from-port="0" to-layer="1912" to-port="1"/>
+ <edge from-layer="1912" from-port="2" to-layer="1914" to-port="0"/>
+ <edge from-layer="1913" from-port="0" to-layer="1914" to-port="1"/>
+ <edge from-layer="1914" from-port="2" to-layer="1915" to-port="0"/>
+ <edge from-layer="1882" from-port="0" to-layer="1915" to-port="1"/>
+ <edge from-layer="1883" from-port="0" to-layer="1915" to-port="2"/>
+ <edge from-layer="1884" from-port="0" to-layer="1915" to-port="3"/>
+ <edge from-layer="1885" from-port="0" to-layer="1915" to-port="4"/>
+ <edge from-layer="1916" from-port="0" to-layer="1917" to-port="0"/>
+ <edge from-layer="1917" from-port="1" to-layer="1919" to-port="0"/>
+ <edge from-layer="1918" from-port="0" to-layer="1919" to-port="1"/>
+ <edge from-layer="1919" from-port="2" to-layer="1921" to-port="0"/>
+ <edge from-layer="1920" from-port="0" to-layer="1921" to-port="1"/>
+ <edge from-layer="1915" from-port="5" to-layer="1922" to-port="0"/>
+ <edge from-layer="1921" from-port="2" to-layer="1922" to-port="1"/>
+ <edge from-layer="1922" from-port="2" to-layer="1924" to-port="0"/>
+ <edge from-layer="1923" from-port="0" to-layer="1924" to-port="1"/>
+ <edge from-layer="1924" from-port="2" to-layer="1925" to-port="0"/>
+ <edge from-layer="1878" from-port="0" to-layer="1925" to-port="1"/>
+ <edge from-layer="1879" from-port="0" to-layer="1925" to-port="2"/>
+ <edge from-layer="1880" from-port="0" to-layer="1925" to-port="3"/>
+ <edge from-layer="1881" from-port="0" to-layer="1925" to-port="4"/>
+ <edge from-layer="1877" from-port="5" to-layer="1926" to-port="0"/>
+ <edge from-layer="1925" from-port="5" to-layer="1926" to-port="1"/>
+ <edge from-layer="1926" from-port="2" to-layer="1928" to-port="0"/>
+ <edge from-layer="1927" from-port="0" to-layer="1928" to-port="1"/>
+ <edge from-layer="1928" from-port="2" to-layer="1929" to-port="0"/>
+ <edge from-layer="12" from-port="0" to-layer="1929" to-port="1"/>
+ <edge from-layer="13" from-port="0" to-layer="1929" to-port="2"/>
+ <edge from-layer="14" from-port="0" to-layer="1929" to-port="3"/>
+ <edge from-layer="15" from-port="0" to-layer="1929" to-port="4"/>
+ <edge from-layer="1942" from-port="0" to-layer="1943" to-port="0"/>
+ <edge from-layer="1943" from-port="1" to-layer="1945" to-port="0"/>
+ <edge from-layer="1944" from-port="0" to-layer="1945" to-port="1"/>
+ <edge from-layer="1945" from-port="2" to-layer="1947" to-port="0"/>
+ <edge from-layer="1946" from-port="0" to-layer="1947" to-port="1"/>
+ <edge from-layer="1929" from-port="5" to-layer="1948" to-port="0"/>
+ <edge from-layer="1947" from-port="2" to-layer="1948" to-port="1"/>
+ <edge from-layer="1948" from-port="2" to-layer="1950" to-port="0"/>
+ <edge from-layer="1949" from-port="0" to-layer="1950" to-port="1"/>
+ <edge from-layer="1950" from-port="2" to-layer="1952" to-port="0"/>
+ <edge from-layer="1951" from-port="0" to-layer="1952" to-port="1"/>
+ <edge from-layer="1952" from-port="2" to-layer="1953" to-port="0"/>
+ <edge from-layer="1938" from-port="0" to-layer="1953" to-port="1"/>
+ <edge from-layer="1939" from-port="0" to-layer="1953" to-port="2"/>
+ <edge from-layer="1940" from-port="0" to-layer="1953" to-port="3"/>
+ <edge from-layer="1941" from-port="0" to-layer="1953" to-port="4"/>
+ <edge from-layer="1955" from-port="0" to-layer="1956" to-port="0"/>
+ <edge from-layer="1956" from-port="1" to-layer="1958" to-port="0"/>
+ <edge from-layer="1957" from-port="0" to-layer="1958" to-port="1"/>
+ <edge from-layer="1958" from-port="2" to-layer="1960" to-port="0"/>
+ <edge from-layer="1959" from-port="0" to-layer="1960" to-port="1"/>
+ <edge from-layer="1960" from-port="2" to-layer="1961" to-port="0"/>
+ <edge from-layer="1954" from-port="0" to-layer="1961" to-port="1"/>
+ <edge from-layer="1953" from-port="5" to-layer="1962" to-port="0"/>
+ <edge from-layer="1961" from-port="2" to-layer="1962" to-port="1"/>
+ <edge from-layer="1962" from-port="2" to-layer="1964" to-port="0"/>
+ <edge from-layer="1963" from-port="0" to-layer="1964" to-port="1"/>
+ <edge from-layer="1964" from-port="2" to-layer="1966" to-port="0"/>
+ <edge from-layer="1965" from-port="0" to-layer="1966" to-port="1"/>
+ <edge from-layer="1966" from-port="2" to-layer="1967" to-port="0"/>
+ <edge from-layer="1934" from-port="0" to-layer="1967" to-port="1"/>
+ <edge from-layer="1935" from-port="0" to-layer="1967" to-port="2"/>
+ <edge from-layer="1936" from-port="0" to-layer="1967" to-port="3"/>
+ <edge from-layer="1937" from-port="0" to-layer="1967" to-port="4"/>
+ <edge from-layer="1968" from-port="0" to-layer="1969" to-port="0"/>
+ <edge from-layer="1969" from-port="1" to-layer="1971" to-port="0"/>
+ <edge from-layer="1970" from-port="0" to-layer="1971" to-port="1"/>
+ <edge from-layer="1971" from-port="2" to-layer="1973" to-port="0"/>
+ <edge from-layer="1972" from-port="0" to-layer="1973" to-port="1"/>
+ <edge from-layer="1967" from-port="5" to-layer="1974" to-port="0"/>
+ <edge from-layer="1973" from-port="2" to-layer="1974" to-port="1"/>
+ <edge from-layer="1974" from-port="2" to-layer="1976" to-port="0"/>
+ <edge from-layer="1975" from-port="0" to-layer="1976" to-port="1"/>
+ <edge from-layer="1976" from-port="2" to-layer="1977" to-port="0"/>
+ <edge from-layer="1930" from-port="0" to-layer="1977" to-port="1"/>
+ <edge from-layer="1931" from-port="0" to-layer="1977" to-port="2"/>
+ <edge from-layer="1932" from-port="0" to-layer="1977" to-port="3"/>
+ <edge from-layer="1933" from-port="0" to-layer="1977" to-port="4"/>
+ <edge from-layer="1929" from-port="5" to-layer="1978" to-port="0"/>
+ <edge from-layer="1977" from-port="5" to-layer="1978" to-port="1"/>
+ <edge from-layer="1978" from-port="2" to-layer="1980" to-port="0"/>
+ <edge from-layer="1979" from-port="0" to-layer="1980" to-port="1"/>
+ <edge from-layer="1980" from-port="2" to-layer="1981" to-port="0"/>
+ <edge from-layer="8" from-port="0" to-layer="1981" to-port="1"/>
+ <edge from-layer="9" from-port="0" to-layer="1981" to-port="2"/>
+ <edge from-layer="10" from-port="0" to-layer="1981" to-port="3"/>
+ <edge from-layer="11" from-port="0" to-layer="1981" to-port="4"/>
+ <edge from-layer="1994" from-port="0" to-layer="1995" to-port="0"/>
+ <edge from-layer="1995" from-port="1" to-layer="1997" to-port="0"/>
+ <edge from-layer="1996" from-port="0" to-layer="1997" to-port="1"/>
+ <edge from-layer="1997" from-port="2" to-layer="1999" to-port="0"/>
+ <edge from-layer="1998" from-port="0" to-layer="1999" to-port="1"/>
+ <edge from-layer="1981" from-port="5" to-layer="2000" to-port="0"/>
+ <edge from-layer="1999" from-port="2" to-layer="2000" to-port="1"/>
+ <edge from-layer="2000" from-port="2" to-layer="2002" to-port="0"/>
+ <edge from-layer="2001" from-port="0" to-layer="2002" to-port="1"/>
+ <edge from-layer="2002" from-port="2" to-layer="2004" to-port="0"/>
+ <edge from-layer="2003" from-port="0" to-layer="2004" to-port="1"/>
+ <edge from-layer="2004" from-port="2" to-layer="2005" to-port="0"/>
+ <edge from-layer="1990" from-port="0" to-layer="2005" to-port="1"/>
+ <edge from-layer="1991" from-port="0" to-layer="2005" to-port="2"/>
+ <edge from-layer="1992" from-port="0" to-layer="2005" to-port="3"/>
+ <edge from-layer="1993" from-port="0" to-layer="2005" to-port="4"/>
+ <edge from-layer="2007" from-port="0" to-layer="2008" to-port="0"/>
+ <edge from-layer="2008" from-port="1" to-layer="2010" to-port="0"/>
+ <edge from-layer="2009" from-port="0" to-layer="2010" to-port="1"/>
+ <edge from-layer="2010" from-port="2" to-layer="2012" to-port="0"/>
+ <edge from-layer="2011" from-port="0" to-layer="2012" to-port="1"/>
+ <edge from-layer="2012" from-port="2" to-layer="2013" to-port="0"/>
+ <edge from-layer="2006" from-port="0" to-layer="2013" to-port="1"/>
+ <edge from-layer="2005" from-port="5" to-layer="2014" to-port="0"/>
+ <edge from-layer="2013" from-port="2" to-layer="2014" to-port="1"/>
+ <edge from-layer="2014" from-port="2" to-layer="2016" to-port="0"/>
+ <edge from-layer="2015" from-port="0" to-layer="2016" to-port="1"/>
+ <edge from-layer="2016" from-port="2" to-layer="2018" to-port="0"/>
+ <edge from-layer="2017" from-port="0" to-layer="2018" to-port="1"/>
+ <edge from-layer="2018" from-port="2" to-layer="2019" to-port="0"/>
+ <edge from-layer="1986" from-port="0" to-layer="2019" to-port="1"/>
+ <edge from-layer="1987" from-port="0" to-layer="2019" to-port="2"/>
+ <edge from-layer="1988" from-port="0" to-layer="2019" to-port="3"/>
+ <edge from-layer="1989" from-port="0" to-layer="2019" to-port="4"/>
+ <edge from-layer="2020" from-port="0" to-layer="2021" to-port="0"/>
+ <edge from-layer="2021" from-port="1" to-layer="2023" to-port="0"/>
+ <edge from-layer="2022" from-port="0" to-layer="2023" to-port="1"/>
+ <edge from-layer="2023" from-port="2" to-layer="2025" to-port="0"/>
+ <edge from-layer="2024" from-port="0" to-layer="2025" to-port="1"/>
+ <edge from-layer="2019" from-port="5" to-layer="2026" to-port="0"/>
+ <edge from-layer="2025" from-port="2" to-layer="2026" to-port="1"/>
+ <edge from-layer="2026" from-port="2" to-layer="2028" to-port="0"/>
+ <edge from-layer="2027" from-port="0" to-layer="2028" to-port="1"/>
+ <edge from-layer="2028" from-port="2" to-layer="2029" to-port="0"/>
+ <edge from-layer="1982" from-port="0" to-layer="2029" to-port="1"/>
+ <edge from-layer="1983" from-port="0" to-layer="2029" to-port="2"/>
+ <edge from-layer="1984" from-port="0" to-layer="2029" to-port="3"/>
+ <edge from-layer="1985" from-port="0" to-layer="2029" to-port="4"/>
+ <edge from-layer="1981" from-port="5" to-layer="2030" to-port="0"/>
+ <edge from-layer="2029" from-port="5" to-layer="2030" to-port="1"/>
+ <edge from-layer="2030" from-port="2" to-layer="2032" to-port="0"/>
+ <edge from-layer="2031" from-port="0" to-layer="2032" to-port="1"/>
+ <edge from-layer="2032" from-port="2" to-layer="2033" to-port="0"/>
+ <edge from-layer="4" from-port="0" to-layer="2033" to-port="1"/>
+ <edge from-layer="5" from-port="0" to-layer="2033" to-port="2"/>
+ <edge from-layer="6" from-port="0" to-layer="2033" to-port="3"/>
+ <edge from-layer="7" from-port="0" to-layer="2033" to-port="4"/>
+ <edge from-layer="2046" from-port="0" to-layer="2047" to-port="0"/>
+ <edge from-layer="2047" from-port="1" to-layer="2049" to-port="0"/>
+ <edge from-layer="2048" from-port="0" to-layer="2049" to-port="1"/>
+ <edge from-layer="2049" from-port="2" to-layer="2051" to-port="0"/>
+ <edge from-layer="2050" from-port="0" to-layer="2051" to-port="1"/>
+ <edge from-layer="2033" from-port="5" to-layer="2052" to-port="0"/>
+ <edge from-layer="2051" from-port="2" to-layer="2052" to-port="1"/>
+ <edge from-layer="2052" from-port="2" to-layer="2054" to-port="0"/>
+ <edge from-layer="2053" from-port="0" to-layer="2054" to-port="1"/>
+ <edge from-layer="2054" from-port="2" to-layer="2056" to-port="0"/>
+ <edge from-layer="2055" from-port="0" to-layer="2056" to-port="1"/>
+ <edge from-layer="2056" from-port="2" to-layer="2057" to-port="0"/>
+ <edge from-layer="2042" from-port="0" to-layer="2057" to-port="1"/>
+ <edge from-layer="2043" from-port="0" to-layer="2057" to-port="2"/>
+ <edge from-layer="2044" from-port="0" to-layer="2057" to-port="3"/>
+ <edge from-layer="2045" from-port="0" to-layer="2057" to-port="4"/>
+ <edge from-layer="2059" from-port="0" to-layer="2060" to-port="0"/>
+ <edge from-layer="2060" from-port="1" to-layer="2062" to-port="0"/>
+ <edge from-layer="2061" from-port="0" to-layer="2062" to-port="1"/>
+ <edge from-layer="2062" from-port="2" to-layer="2064" to-port="0"/>
+ <edge from-layer="2063" from-port="0" to-layer="2064" to-port="1"/>
+ <edge from-layer="2064" from-port="2" to-layer="2065" to-port="0"/>
+ <edge from-layer="2058" from-port="0" to-layer="2065" to-port="1"/>
+ <edge from-layer="2057" from-port="5" to-layer="2066" to-port="0"/>
+ <edge from-layer="2065" from-port="2" to-layer="2066" to-port="1"/>
+ <edge from-layer="2066" from-port="2" to-layer="2068" to-port="0"/>
+ <edge from-layer="2067" from-port="0" to-layer="2068" to-port="1"/>
+ <edge from-layer="2068" from-port="2" to-layer="2070" to-port="0"/>
+ <edge from-layer="2069" from-port="0" to-layer="2070" to-port="1"/>
+ <edge from-layer="2070" from-port="2" to-layer="2071" to-port="0"/>
+ <edge from-layer="2038" from-port="0" to-layer="2071" to-port="1"/>
+ <edge from-layer="2039" from-port="0" to-layer="2071" to-port="2"/>
+ <edge from-layer="2040" from-port="0" to-layer="2071" to-port="3"/>
+ <edge from-layer="2041" from-port="0" to-layer="2071" to-port="4"/>
+ <edge from-layer="2072" from-port="0" to-layer="2073" to-port="0"/>
+ <edge from-layer="2073" from-port="1" to-layer="2075" to-port="0"/>
+ <edge from-layer="2074" from-port="0" to-layer="2075" to-port="1"/>
+ <edge from-layer="2075" from-port="2" to-layer="2077" to-port="0"/>
+ <edge from-layer="2076" from-port="0" to-layer="2077" to-port="1"/>
+ <edge from-layer="2071" from-port="5" to-layer="2078" to-port="0"/>
+ <edge from-layer="2077" from-port="2" to-layer="2078" to-port="1"/>
+ <edge from-layer="2078" from-port="2" to-layer="2080" to-port="0"/>
+ <edge from-layer="2079" from-port="0" to-layer="2080" to-port="1"/>
+ <edge from-layer="2080" from-port="2" to-layer="2081" to-port="0"/>
+ <edge from-layer="2034" from-port="0" to-layer="2081" to-port="1"/>
+ <edge from-layer="2035" from-port="0" to-layer="2081" to-port="2"/>
+ <edge from-layer="2036" from-port="0" to-layer="2081" to-port="3"/>
+ <edge from-layer="2037" from-port="0" to-layer="2081" to-port="4"/>
+ <edge from-layer="2033" from-port="5" to-layer="2082" to-port="0"/>
+ <edge from-layer="2081" from-port="5" to-layer="2082" to-port="1"/>
+ <edge from-layer="2082" from-port="2" to-layer="2084" to-port="0"/>
+ <edge from-layer="2083" from-port="0" to-layer="2084" to-port="1"/>
+ <edge from-layer="2084" from-port="2" to-layer="2085" to-port="0"/>
+ <edge from-layer="0" from-port="0" to-layer="2085" to-port="1"/>
+ <edge from-layer="1" from-port="0" to-layer="2085" to-port="2"/>
+ <edge from-layer="2" from-port="0" to-layer="2085" to-port="3"/>
+ <edge from-layer="3" from-port="0" to-layer="2085" to-port="4"/>
+ <edge from-layer="2086" from-port="0" to-layer="2087" to-port="0"/>
+ <edge from-layer="2087" from-port="1" to-layer="2089" to-port="0"/>
+ <edge from-layer="2088" from-port="0" to-layer="2089" to-port="1"/>
+ <edge from-layer="2089" from-port="2" to-layer="2091" to-port="0"/>
+ <edge from-layer="2090" from-port="0" to-layer="2091" to-port="1"/>
+ <edge from-layer="2085" from-port="5" to-layer="2092" to-port="0"/>
+ <edge from-layer="2091" from-port="2" to-layer="2092" to-port="1"/>
+ <edge from-layer="2092" from-port="2" to-layer="2094" to-port="0"/>
+ <edge from-layer="2093" from-port="0" to-layer="2094" to-port="1"/>
+ <edge from-layer="2094" from-port="2" to-layer="2096" to-port="0"/>
+ <edge from-layer="2095" from-port="0" to-layer="2096" to-port="1"/>
+ <edge from-layer="2096" from-port="2" to-layer="2098" to-port="0"/>
+ <edge from-layer="2097" from-port="0" to-layer="2098" to-port="1"/>
+ <edge from-layer="2099" from-port="0" to-layer="2100" to-port="0"/>
+ <edge from-layer="2100" from-port="1" to-layer="2102" to-port="0"/>
+ <edge from-layer="2101" from-port="0" to-layer="2102" to-port="1"/>
+ <edge from-layer="2102" from-port="2" to-layer="2104" to-port="0"/>
+ <edge from-layer="2103" from-port="0" to-layer="2104" to-port="1"/>
+ <edge from-layer="2085" from-port="5" to-layer="2105" to-port="0"/>
+ <edge from-layer="2104" from-port="2" to-layer="2105" to-port="1"/>
+ <edge from-layer="2105" from-port="2" to-layer="2107" to-port="0"/>
+ <edge from-layer="2106" from-port="0" to-layer="2107" to-port="1"/>
+ <edge from-layer="2107" from-port="2" to-layer="2109" to-port="0"/>
+ <edge from-layer="2108" from-port="0" to-layer="2109" to-port="1"/>
+ <edge from-layer="2109" from-port="2" to-layer="2111" to-port="0"/>
+ <edge from-layer="2110" from-port="0" to-layer="2111" to-port="1"/>
+ <edge from-layer="2111" from-port="2" to-layer="2113" to-port="0"/>
+ <edge from-layer="2112" from-port="0" to-layer="2113" to-port="1"/>
+ <edge from-layer="2113" from-port="2" to-layer="2114" to-port="0"/>
+ <edge from-layer="2114" from-port="1" to-layer="2116" to-port="0"/>
+ <edge from-layer="2115" from-port="0" to-layer="2116" to-port="1"/>
+ <edge from-layer="2084" from-port="2" to-layer="2117" to-port="0"/>
+ <edge from-layer="2117" from-port="1" to-layer="2121" to-port="0"/>
+ <edge from-layer="2118" from-port="0" to-layer="2121" to-port="1"/>
+ <edge from-layer="2119" from-port="0" to-layer="2121" to-port="2"/>
+ <edge from-layer="2120" from-port="0" to-layer="2121" to-port="3"/>
+ <edge from-layer="164" from-port="0" to-layer="2122" to-port="0"/>
+ <edge from-layer="2122" from-port="1" to-layer="2126" to-port="0"/>
+ <edge from-layer="2123" from-port="0" to-layer="2126" to-port="1"/>
+ <edge from-layer="2124" from-port="0" to-layer="2126" to-port="2"/>
+ <edge from-layer="2125" from-port="0" to-layer="2126" to-port="3"/>
+ <edge from-layer="2121" from-port="4" to-layer="2127" to-port="0"/>
+ <edge from-layer="2126" from-port="4" to-layer="2127" to-port="1"/>
+ <edge from-layer="2127" from-port="2" to-layer="2129" to-port="0"/>
+ <edge from-layer="2128" from-port="0" to-layer="2129" to-port="1"/>
+ <edge from-layer="2098" from-port="2" to-layer="2130" to-port="0"/>
+ <edge from-layer="2116" from-port="2" to-layer="2130" to-port="1"/>
+ <edge from-layer="2129" from-port="2" to-layer="2130" to-port="2"/>
+ <edge from-layer="2130" from-port="3" to-layer="2131" to-port="0"/>
+ </edges>
+ <meta_data>
+ <MO_version value="custom_HEAD_149c43044cb1e8ed8cd4f3f196b23f7b3f129a36"/>
+ <cli_parameters>
+ <caffe_parser_path value="DIR"/>
+ <data_type value="FP16"/>
+ <disable_nhwc_to_nchw value="False"/>
+ <disable_omitting_optional value="False"/>
+ <disable_resnet_optimization value="False"/>
+ <disable_weights_compression value="False"/>
+ <enable_concat_optimization value="False"/>
+ <enable_flattening_nested_params value="False"/>
+ <enable_ssd_gluoncv value="False"/>
+ <extensions value="DIR"/>
+ <framework value="caffe"/>
+ <freeze_placeholder_with_value value="{}"/>
+ <generate_deprecated_IR_V7 value="False"/>
+ <input value="data"/>
+ <input_model value="DIR/rmnet_lrelu_pd_ssd.caffemodel"/>
+ <input_model_is_text value="False"/>
+ <input_proto value="DIR/rmnet_lrelu_pd_ssd.prototxt"/>
+ <input_shape value="[1,3,320,544]"/>
+ <k value="DIR/CustomLayersMapping.xml"/>
+ <keep_shape_ops value="True"/>
+ <legacy_ir_generation value="False"/>
+ <legacy_mxnet_model value="False"/>
+ <log_level value="ERROR"/>
+ <mean_scale_values value="{}"/>
+ <mean_values value="()"/>
+ <model_name value="person-detection-retail-0013"/>
+ <output value="['detection_out']"/>
+ <output_dir value="DIR"/>
+ <placeholder_data_types value="{}"/>
+ <placeholder_shapes value="{'data': array([ 1, 3, 320, 544])}"/>
+ <progress value="False"/>
+ <remove_memory value="False"/>
+ <remove_output_softmax value="False"/>
+ <reverse_input_channels value="False"/>
+ <save_params_from_nd value="False"/>
+ <scale_values value="()"/>
+ <silent value="False"/>
+ <static_shape value="False"/>
+ <stream_output value="False"/>
+ <transform value=""/>
+ <unset unset_cli_parameters="batch, counts, disable_fusing, disable_gfusing, finegrain_fusing, input_checkpoint, input_meta_graph, input_symbol, mean_file, mean_file_offsets, move_to_preprocess, nd_prefix_name, pretrained_model_name, saved_model_dir, saved_model_tags, scale, tensorboard_logdir, tensorflow_custom_layer_libraries, tensorflow_custom_operations_config_update, tensorflow_object_detection_api_pipeline_config, tensorflow_use_custom_operations_config, transformations_config"/>
+ </cli_parameters>
+ </meta_data>
+ <quantization_parameters>
+ <config>{
+ 'compression': {
+ 'algorithms': [
+ {
+ 'name': 'DefaultQuantization',
+ 'params': {
+ 'num_samples_for_tuning': 2000,
+ 'preset': 'performance',
+ 'stat_subset_size': 300,
+ 'use_layerwise_tuning': false
+ }
+ }
+ ],
+ 'dump_intermediate_model': true,
+ 'target_device': 'ANY'
+ },
+ 'engine': {
+ 'models': [
+ {
+ 'name': 'person-detection-retail-0013',
+ 'launchers': [
+ {
+ 'framework': 'dlsdk',
+ 'adapter': 'ssd',
+ 'device': 'CPU'
+ }
+ ],
+ 'datasets': [
+ {
+ 'name': 'person_detection',
+ 'data_source': 'PATH',
+ 'annotation_conversion': {
+ 'converter': 'datatang_street_subway',
+ 'annotation_file': 'PATH'
+ },
+ 'annotation': 'PATH',
+ 'dataset_meta': 'PATH',
+ 'preprocessing': [
+ {
+ 'type': 'resize',
+ 'dst_width': 544,
+ 'dst_height': 320
+ }
+ ],
+ 'postprocessing': [
+ {
+ 'type': 'resize_prediction_boxes'
+ },
+ {
+ 'type': 'filter',
+ 'apply_to': 'annotation',
+ 'height_range': 100,
+ 'aspect_ratio': '0.666, 5',
+ 'is_empty': true,
+ 'min_visibility': 'partially occluded'
+ },
+ {
+ 'type': 'filter',
+ 'apply_to': 'prediction',
+ 'height_range': 100,
+ 'is_empty': true,
+ 'aspect_ratio': '0.666, 5'
+ }
+ ],
+ 'metrics': [
+ {
+ 'type': 'map',
+ 'ignore_difficult': true,
+ 'include_boundaries': true,
+ 'allow_multiple_matches_per_ignored': false,
+ 'distinct_conf': false
+ }
+ ],
+ '_command_line_mapping': {
+ 'annotation_file': 'PATH'
+ }
+ }
+ ]
+ }
+ ],
+ 'stat_requests_number': null,
+ 'eval_requests_number': null,
+ 'type': 'accuracy_checker'
+ }
+ }</config>
+ <version value="1.0"/>
+ <cli_params value="{'quantize': None, 'preset': None, 'model': None, 'weights': None, 'name': None, 'ac_config': None, 'max_drop': None, 'evaluate': False, 'output_dir': 'PATH', 'direct_dump': True, 'log_level': 'INFO', 'pbar': False, 'stream_output': False, 'keep_uncompressed_weights': False}"/>
+ </quantization_parameters>
+</net>
diff --git a/resources/models/intel/person-detection-retail-0013/FP16/person-detection-retail-0013.bin b/resources/models/intel/person-detection-retail-0013/FP16/person-detection-retail-0013.bin
new file mode 100644
index 0000000..ddc38c5
--- /dev/null
+++ b/resources/models/intel/person-detection-retail-0013/FP16/person-detection-retail-0013.bin
Binary files differ
diff --git a/resources/models/intel/person-detection-retail-0013/FP16/person-detection-retail-0013.xml b/resources/models/intel/person-detection-retail-0013/FP16/person-detection-retail-0013.xml
new file mode 100644
index 0000000..2d45fca
--- /dev/null
+++ b/resources/models/intel/person-detection-retail-0013/FP16/person-detection-retail-0013.xml
@@ -0,0 +1,13694 @@
+<?xml version="1.0" ?>
+<net name="ResMobNet_v4 (LReLU) with single SSD head" version="10">
+ <layers>
+ <layer id="0" name="data" type="Parameter" version="opset1">
+ <data element_type="f16" shape="1, 3, 320, 544"/>
+ <output>
+ <port id="0" names="data" precision="FP16">
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>320</dim>
+ <dim>544</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1" name="data_mul_23644" type="Const" version="opset1">
+ <data element_type="f16" offset="0" shape="1, 3, 1, 1" size="6"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="2" name="data/norm/bn/mean/Fused_Mul_" type="Multiply" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>320</dim>
+ <dim>544</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>320</dim>
+ <dim>544</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="3" name="data_add_23646" type="Const" version="opset1">
+ <data element_type="f16" offset="6" shape="1, 3, 1, 1" size="6"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="4" name="data/norm/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>320</dim>
+ <dim>544</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="data/norm/bn" precision="FP16">
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>320</dim>
+ <dim>544</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="5" name="init_block1/dim_inc/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f16" offset="12" shape="32, 3, 3, 3" size="1728"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>32</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="6" name="init_block1/dim_inc/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="1, 1" pads_end="1, 1" strides="2, 2"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>320</dim>
+ <dim>544</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>32</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="7" name="data_add_2364923654" type="Const" version="opset1">
+ <data element_type="f16" offset="1740" shape="1, 32, 1, 1" size="64"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="8" name="init_block1/dim_inc/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="init_block1/dim_inc/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="9" name="init_block1/dim_inc/fn" type="ReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ </input>
+ <output>
+ <port id="1" names="init_block1/dim_inc/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="10" name="bottleneck1_1/dim_red/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f16" offset="1804" shape="8, 32, 1, 1" size="512"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>8</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="11" name="bottleneck1_1/dim_red/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="0, 0" pads_end="0, 0" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>8</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="12" name="data_add_2365723662" type="Const" version="opset1">
+ <data element_type="f16" offset="2316" shape="1, 8, 1, 1" size="16"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="13" name="bottleneck1_1/dim_red/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck1_1/dim_red/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="14" name="bottleneck1_1/dim_red/fn/weights3098439866" type="Const" version="opset1">
+ <data element_type="f32" offset="2332" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="15" name="bottleneck1_1/dim_red/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck1_1/dim_red/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="16" name="bottleneck1_1/inner/dw1/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f16" offset="2336" shape="8, 1, 1, 3, 3" size="144"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="17" name="bottleneck1_1/inner/dw1/conv" type="GroupConvolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="1, 1" pads_end="1, 1" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="18" name="data_add_2366523670" type="Const" version="opset1">
+ <data element_type="f16" offset="2480" shape="1, 8, 1, 1" size="16"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="19" name="bottleneck1_1/inner/dw1/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck1_1/inner/dw1/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="20" name="bottleneck1_1/inner/dw1/fn/weights3117639995" type="Const" version="opset1">
+ <data element_type="f32" offset="2332" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="21" name="bottleneck1_1/inner/dw1/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck1_1/inner/dw1/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="22" name="bottleneck1_1/dim_inc/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f16" offset="2496" shape="32, 8, 1, 1" size="512"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>32</dim>
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="23" name="bottleneck1_1/dim_inc/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="0, 0" pads_end="0, 0" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>32</dim>
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="24" name="data_add_2367323678" type="Const" version="opset1">
+ <data element_type="f16" offset="3008" shape="1, 32, 1, 1" size="64"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="25" name="bottleneck1_1/dim_inc/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck1_1/dim_inc/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="26" name="bottleneck1_1/add" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck1_1/add" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="27" name="bottleneck1_1/fn/weights3117240232" type="Const" version="opset1">
+ <data element_type="f32" offset="2332" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="28" name="bottleneck1_1/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck1_1/add" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="29" name="bottleneck1_2/dim_red/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f16" offset="3072" shape="8, 32, 1, 1" size="512"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>8</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="30" name="bottleneck1_2/dim_red/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="0, 0" pads_end="0, 0" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>8</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="31" name="data_add_2368123686" type="Const" version="opset1">
+ <data element_type="f16" offset="3584" shape="1, 8, 1, 1" size="16"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="32" name="bottleneck1_2/dim_red/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck1_2/dim_red/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="33" name="bottleneck1_2/dim_red/fn/weights3112040508" type="Const" version="opset1">
+ <data element_type="f32" offset="2332" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="34" name="bottleneck1_2/dim_red/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck1_2/dim_red/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="35" name="bottleneck1_2/inner/dw1/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f16" offset="3600" shape="8, 1, 1, 3, 3" size="144"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="36" name="bottleneck1_2/inner/dw1/conv" type="GroupConvolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="1, 1" pads_end="1, 1" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="37" name="data_add_2368923694" type="Const" version="opset1">
+ <data element_type="f16" offset="3744" shape="1, 8, 1, 1" size="16"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="38" name="bottleneck1_2/inner/dw1/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck1_2/inner/dw1/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="39" name="bottleneck1_2/inner/dw1/fn/weights3080440244" type="Const" version="opset1">
+ <data element_type="f32" offset="2332" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="40" name="bottleneck1_2/inner/dw1/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck1_2/inner/dw1/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="41" name="bottleneck1_2/dim_inc/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f16" offset="3760" shape="32, 8, 1, 1" size="512"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>32</dim>
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="42" name="bottleneck1_2/dim_inc/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="0, 0" pads_end="0, 0" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>32</dim>
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="43" name="data_add_2369723702" type="Const" version="opset1">
+ <data element_type="f16" offset="4272" shape="1, 32, 1, 1" size="64"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="44" name="bottleneck1_2/dim_inc/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck1_2/dim_inc/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="45" name="bottleneck1_2/add" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck1_2/add" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="46" name="bottleneck1_2/fn/weights3113640172" type="Const" version="opset1">
+ <data element_type="f32" offset="2332" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="47" name="bottleneck1_2/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck1_2/add" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="48" name="bottleneck1_3/dim_red/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f16" offset="4336" shape="8, 32, 1, 1" size="512"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>8</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="49" name="bottleneck1_3/dim_red/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="0, 0" pads_end="0, 0" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>8</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="50" name="data_add_2370523710" type="Const" version="opset1">
+ <data element_type="f16" offset="4848" shape="1, 8, 1, 1" size="16"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="51" name="bottleneck1_3/dim_red/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck1_3/dim_red/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="52" name="bottleneck1_3/dim_red/fn/weights3091240304" type="Const" version="opset1">
+ <data element_type="f32" offset="2332" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="53" name="bottleneck1_3/dim_red/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck1_3/dim_red/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="54" name="bottleneck1_3/inner/dw1/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f16" offset="4864" shape="8, 1, 1, 3, 3" size="144"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="55" name="bottleneck1_3/inner/dw1/conv" type="GroupConvolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="1, 1" pads_end="1, 1" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="56" name="data_add_2371323718" type="Const" version="opset1">
+ <data element_type="f16" offset="5008" shape="1, 8, 1, 1" size="16"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="57" name="bottleneck1_3/inner/dw1/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck1_3/inner/dw1/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="58" name="bottleneck1_3/inner/dw1/fn/weights3110039956" type="Const" version="opset1">
+ <data element_type="f32" offset="2332" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="59" name="bottleneck1_3/inner/dw1/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck1_3/inner/dw1/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="60" name="bottleneck1_3/dim_inc/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f16" offset="5024" shape="32, 8, 1, 1" size="512"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>32</dim>
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="61" name="bottleneck1_3/dim_inc/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="0, 0" pads_end="0, 0" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>32</dim>
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="62" name="data_add_2372123726" type="Const" version="opset1">
+ <data element_type="f16" offset="5536" shape="1, 32, 1, 1" size="64"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="63" name="bottleneck1_3/dim_inc/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck1_3/dim_inc/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="64" name="bottleneck1_3/add" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck1_3/add" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="65" name="bottleneck1_3/fn/weights3091640319" type="Const" version="opset1">
+ <data element_type="f32" offset="2332" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="66" name="bottleneck1_3/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck1_3/add" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="67" name="bottleneck1_4/dim_red/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f16" offset="5600" shape="8, 32, 1, 1" size="512"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>8</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="68" name="bottleneck1_4/dim_red/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="0, 0" pads_end="0, 0" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>8</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="69" name="data_add_2372923734" type="Const" version="opset1">
+ <data element_type="f16" offset="6112" shape="1, 8, 1, 1" size="16"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="70" name="bottleneck1_4/dim_red/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck1_4/dim_red/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="71" name="bottleneck1_4/dim_red/fn/weights3110439965" type="Const" version="opset1">
+ <data element_type="f32" offset="2332" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="72" name="bottleneck1_4/dim_red/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck1_4/dim_red/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="73" name="bottleneck1_4/inner/dw1/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f16" offset="6128" shape="8, 1, 1, 3, 3" size="144"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="74" name="bottleneck1_4/inner/dw1/conv" type="GroupConvolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="1, 1" pads_end="1, 1" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="75" name="data_add_2373723742" type="Const" version="opset1">
+ <data element_type="f16" offset="6272" shape="1, 8, 1, 1" size="16"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="76" name="bottleneck1_4/inner/dw1/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck1_4/inner/dw1/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="77" name="bottleneck1_4/inner/dw1/fn/weights3102040547" type="Const" version="opset1">
+ <data element_type="f32" offset="2332" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="78" name="bottleneck1_4/inner/dw1/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck1_4/inner/dw1/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="79" name="bottleneck1_4/dim_inc/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f16" offset="6288" shape="32, 8, 1, 1" size="512"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>32</dim>
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="80" name="bottleneck1_4/dim_inc/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="0, 0" pads_end="0, 0" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>32</dim>
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="81" name="data_add_2374523750" type="Const" version="opset1">
+ <data element_type="f16" offset="6800" shape="1, 32, 1, 1" size="64"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="82" name="bottleneck1_4/dim_inc/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck1_4/dim_inc/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="83" name="bottleneck1_4/add" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck1_4/add" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="84" name="bottleneck1_4/fn/weights3094440088" type="Const" version="opset1">
+ <data element_type="f32" offset="2332" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="85" name="bottleneck1_4/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck1_4/add" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="86" name="bottleneck2_0/skip/pooling" type="MaxPool" version="opset1">
+ <data auto_pad="explicit" kernel="2, 2" pads_begin="0, 0" pads_end="0, 0" rounding_type="ceil" strides="2, 2"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ </input>
+ <output>
+ <port id="1" names="bottleneck2_0/skip/pooling" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="87" name="bottleneck2_0/skip/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f16" offset="6864" shape="64, 32, 1, 1" size="4096"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>64</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="88" name="bottleneck2_0/skip/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="0, 0" pads_end="0, 0" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>64</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="89" name="data_add_2375323758" type="Const" version="opset1">
+ <data element_type="f16" offset="10960" shape="1, 64, 1, 1" size="128"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="90" name="bottleneck2_0/skip/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck2_0/skip/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="91" name="bottleneck2_0/dim_red/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f16" offset="11088" shape="16, 32, 1, 1" size="1024"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>16</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="92" name="bottleneck2_0/dim_red/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="0, 0" pads_end="0, 0" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>16</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="93" name="data_add_2376123766" type="Const" version="opset1">
+ <data element_type="f16" offset="12112" shape="1, 16, 1, 1" size="32"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="94" name="bottleneck2_0/dim_red/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck2_0/dim_red/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="95" name="bottleneck2_0/dim_red/fn/weights3105640580" type="Const" version="opset1">
+ <data element_type="f32" offset="2332" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="96" name="bottleneck2_0/dim_red/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck2_0/dim_red/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="97" name="bottleneck2_0/inner/dw1/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f16" offset="12144" shape="16, 1, 1, 3, 3" size="288"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="98" name="bottleneck2_0/inner/dw1/conv" type="GroupConvolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="1, 1" pads_end="1, 1" strides="2, 2"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="99" name="data_add_2376923774" type="Const" version="opset1">
+ <data element_type="f16" offset="12432" shape="1, 16, 1, 1" size="32"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="100" name="bottleneck2_0/inner/dw1/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck2_0/inner/dw1/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="101" name="bottleneck2_0/inner/dw1/fn/weights3113240199" type="Const" version="opset1">
+ <data element_type="f32" offset="2332" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="102" name="bottleneck2_0/inner/dw1/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck2_0/inner/dw1/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="103" name="bottleneck2_0/dim_inc/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f16" offset="12464" shape="64, 16, 1, 1" size="2048"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>64</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="104" name="bottleneck2_0/dim_inc/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="0, 0" pads_end="0, 0" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>64</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="105" name="data_add_2377723782" type="Const" version="opset1">
+ <data element_type="f16" offset="14512" shape="1, 64, 1, 1" size="128"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="106" name="bottleneck2_0/dim_inc/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck2_0/dim_inc/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="107" name="bottleneck2_0/add" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck2_0/add" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="108" name="bottleneck2_0/fn/weights3100839908" type="Const" version="opset1">
+ <data element_type="f32" offset="2332" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="109" name="bottleneck2_0/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck2_0/add" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="110" name="bottleneck2_1/dim_red/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f16" offset="14640" shape="16, 64, 1, 1" size="2048"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>16</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="111" name="bottleneck2_1/dim_red/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="0, 0" pads_end="0, 0" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>16</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="112" name="data_add_2378523790" type="Const" version="opset1">
+ <data element_type="f16" offset="16688" shape="1, 16, 1, 1" size="32"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="113" name="bottleneck2_1/dim_red/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck2_1/dim_red/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="114" name="bottleneck2_1/dim_red/fn/weights3103240019" type="Const" version="opset1">
+ <data element_type="f32" offset="2332" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="115" name="bottleneck2_1/dim_red/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck2_1/dim_red/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="116" name="bottleneck2_1/inner/dw1/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f16" offset="16720" shape="16, 1, 1, 3, 3" size="288"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="117" name="bottleneck2_1/inner/dw1/conv" type="GroupConvolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="1, 1" pads_end="1, 1" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="118" name="data_add_2379323798" type="Const" version="opset1">
+ <data element_type="f16" offset="17008" shape="1, 16, 1, 1" size="32"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="119" name="bottleneck2_1/inner/dw1/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck2_1/inner/dw1/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="120" name="bottleneck2_1/inner/dw1/fn/weights3092039812" type="Const" version="opset1">
+ <data element_type="f32" offset="2332" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="121" name="bottleneck2_1/inner/dw1/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck2_1/inner/dw1/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="122" name="bottleneck2_1/dim_inc/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f16" offset="17040" shape="64, 16, 1, 1" size="2048"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>64</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="123" name="bottleneck2_1/dim_inc/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="0, 0" pads_end="0, 0" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>64</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="124" name="data_add_2380123806" type="Const" version="opset1">
+ <data element_type="f16" offset="19088" shape="1, 64, 1, 1" size="128"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="125" name="bottleneck2_1/dim_inc/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck2_1/dim_inc/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="126" name="bottleneck2_1/add" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck2_1/add" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="127" name="bottleneck2_1/fn/weights3102840289" type="Const" version="opset1">
+ <data element_type="f32" offset="2332" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="128" name="bottleneck2_1/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck2_1/add" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="129" name="bottleneck2_2/dim_red/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f16" offset="19216" shape="16, 64, 1, 1" size="2048"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>16</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="130" name="bottleneck2_2/dim_red/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="0, 0" pads_end="0, 0" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>16</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="131" name="data_add_2380923814" type="Const" version="opset1">
+ <data element_type="f16" offset="21264" shape="1, 16, 1, 1" size="32"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="132" name="bottleneck2_2/dim_red/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck2_2/dim_red/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="133" name="bottleneck2_2/dim_red/fn/weights3090439815" type="Const" version="opset1">
+ <data element_type="f32" offset="2332" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="134" name="bottleneck2_2/dim_red/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck2_2/dim_red/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="135" name="bottleneck2_2/inner/dw1/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f16" offset="21296" shape="16, 1, 1, 3, 3" size="288"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="136" name="bottleneck2_2/inner/dw1/conv" type="GroupConvolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="1, 1" pads_end="1, 1" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="137" name="data_add_2381723822" type="Const" version="opset1">
+ <data element_type="f16" offset="21584" shape="1, 16, 1, 1" size="32"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="138" name="bottleneck2_2/inner/dw1/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck2_2/inner/dw1/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="139" name="bottleneck2_2/inner/dw1/fn/weights3105240634" type="Const" version="opset1">
+ <data element_type="f32" offset="2332" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="140" name="bottleneck2_2/inner/dw1/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck2_2/inner/dw1/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="141" name="bottleneck2_2/dim_inc/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f16" offset="21616" shape="64, 16, 1, 1" size="2048"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>64</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="142" name="bottleneck2_2/dim_inc/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="0, 0" pads_end="0, 0" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>64</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="143" name="data_add_2382523830" type="Const" version="opset1">
+ <data element_type="f16" offset="23664" shape="1, 64, 1, 1" size="128"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="144" name="bottleneck2_2/dim_inc/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck2_2/dim_inc/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="145" name="bottleneck2_2/add" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck2_2/add" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="146" name="bottleneck2_2/fn/weights3115240121" type="Const" version="opset1">
+ <data element_type="f32" offset="2332" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="147" name="bottleneck2_2/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck2_2/add" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="148" name="bottleneck2_3/dim_red/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f16" offset="23792" shape="16, 64, 1, 1" size="2048"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>16</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="149" name="bottleneck2_3/dim_red/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="0, 0" pads_end="0, 0" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>16</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="150" name="data_add_2383323838" type="Const" version="opset1">
+ <data element_type="f16" offset="25840" shape="1, 16, 1, 1" size="32"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="151" name="bottleneck2_3/dim_red/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck2_3/dim_red/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="152" name="bottleneck2_3/dim_red/fn/weights3114840157" type="Const" version="opset1">
+ <data element_type="f32" offset="2332" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="153" name="bottleneck2_3/dim_red/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck2_3/dim_red/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="154" name="bottleneck2_3/inner/dw1/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f16" offset="25872" shape="16, 1, 1, 3, 3" size="288"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="155" name="bottleneck2_3/inner/dw1/conv" type="GroupConvolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="1, 1" pads_end="1, 1" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="156" name="data_add_2384123846" type="Const" version="opset1">
+ <data element_type="f16" offset="26160" shape="1, 16, 1, 1" size="32"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="157" name="bottleneck2_3/inner/dw1/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck2_3/inner/dw1/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="158" name="bottleneck2_3/inner/dw1/fn/weights3111640571" type="Const" version="opset1">
+ <data element_type="f32" offset="2332" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="159" name="bottleneck2_3/inner/dw1/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck2_3/inner/dw1/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="160" name="bottleneck2_3/dim_inc/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f16" offset="26192" shape="64, 16, 1, 1" size="2048"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>64</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="161" name="bottleneck2_3/dim_inc/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="0, 0" pads_end="0, 0" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>64</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="162" name="data_add_2384923854" type="Const" version="opset1">
+ <data element_type="f16" offset="28240" shape="1, 64, 1, 1" size="128"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="163" name="bottleneck2_3/dim_inc/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck2_3/dim_inc/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="164" name="bottleneck2_3/add" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck2_3/add" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="165" name="bottleneck2_3/fn/weights3106839797" type="Const" version="opset1">
+ <data element_type="f32" offset="2332" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="166" name="bottleneck2_3/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck2_3/add" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="167" name="bottleneck2_4/dim_red/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f16" offset="28368" shape="16, 64, 1, 1" size="2048"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>16</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="168" name="bottleneck2_4/dim_red/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="0, 0" pads_end="0, 0" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>16</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="169" name="data_add_2385723862" type="Const" version="opset1">
+ <data element_type="f16" offset="30416" shape="1, 16, 1, 1" size="32"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="170" name="bottleneck2_4/dim_red/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck2_4/dim_red/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="171" name="bottleneck2_4/dim_red/fn/weights3101240241" type="Const" version="opset1">
+ <data element_type="f32" offset="2332" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="172" name="bottleneck2_4/dim_red/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck2_4/dim_red/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="173" name="bottleneck2_4/inner/dw1/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f16" offset="30448" shape="16, 1, 1, 3, 3" size="288"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="174" name="bottleneck2_4/inner/dw1/conv" type="GroupConvolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="1, 1" pads_end="1, 1" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="175" name="data_add_2386523870" type="Const" version="opset1">
+ <data element_type="f16" offset="30736" shape="1, 16, 1, 1" size="32"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="176" name="bottleneck2_4/inner/dw1/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck2_4/inner/dw1/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="177" name="bottleneck2_4/inner/dw1/fn/weights3116840295" type="Const" version="opset1">
+ <data element_type="f32" offset="2332" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="178" name="bottleneck2_4/inner/dw1/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck2_4/inner/dw1/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="179" name="bottleneck2_4/dim_inc/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f16" offset="30768" shape="64, 16, 1, 1" size="2048"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>64</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="180" name="bottleneck2_4/dim_inc/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="0, 0" pads_end="0, 0" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>64</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="181" name="data_add_2387323878" type="Const" version="opset1">
+ <data element_type="f16" offset="32816" shape="1, 64, 1, 1" size="128"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="182" name="bottleneck2_4/dim_inc/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck2_4/dim_inc/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="183" name="bottleneck2_4/add" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck2_4/add" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="184" name="bottleneck2_4/fn/weights3076440628" type="Const" version="opset1">
+ <data element_type="f32" offset="2332" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="185" name="bottleneck2_4/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck2_4/add" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="186" name="bottleneck2_5/dim_red/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f16" offset="32944" shape="16, 64, 1, 1" size="2048"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>16</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="187" name="bottleneck2_5/dim_red/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="0, 0" pads_end="0, 0" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>16</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="188" name="data_add_2388123886" type="Const" version="opset1">
+ <data element_type="f16" offset="34992" shape="1, 16, 1, 1" size="32"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="189" name="bottleneck2_5/dim_red/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck2_5/dim_red/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="190" name="bottleneck2_5/dim_red/fn/weights3098840499" type="Const" version="opset1">
+ <data element_type="f32" offset="2332" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="191" name="bottleneck2_5/dim_red/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck2_5/dim_red/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="192" name="bottleneck2_5/inner/dw1/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f16" offset="35024" shape="16, 1, 1, 3, 3" size="288"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="193" name="bottleneck2_5/inner/dw1/conv" type="GroupConvolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="1, 1" pads_end="1, 1" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="194" name="data_add_2388923894" type="Const" version="opset1">
+ <data element_type="f16" offset="35312" shape="1, 16, 1, 1" size="32"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="195" name="bottleneck2_5/inner/dw1/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck2_5/inner/dw1/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="196" name="bottleneck2_5/inner/dw1/fn/weights3108040145" type="Const" version="opset1">
+ <data element_type="f32" offset="2332" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="197" name="bottleneck2_5/inner/dw1/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck2_5/inner/dw1/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="198" name="bottleneck2_5/dim_inc/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f16" offset="35344" shape="64, 16, 1, 1" size="2048"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>64</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="199" name="bottleneck2_5/dim_inc/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="0, 0" pads_end="0, 0" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>64</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="200" name="data_add_2389723902" type="Const" version="opset1">
+ <data element_type="f16" offset="37392" shape="1, 64, 1, 1" size="128"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="201" name="bottleneck2_5/dim_inc/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck2_5/dim_inc/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="202" name="bottleneck2_5/add" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck2_5/add" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="203" name="bottleneck2_5/fn/weights3076840217" type="Const" version="opset1">
+ <data element_type="f32" offset="2332" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="204" name="bottleneck2_5/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck2_5/add" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="205" name="bottleneck2_6/dim_red/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f16" offset="37520" shape="16, 64, 1, 1" size="2048"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>16</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="206" name="bottleneck2_6/dim_red/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="0, 0" pads_end="0, 0" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>16</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="207" name="data_add_2390523910" type="Const" version="opset1">
+ <data element_type="f16" offset="39568" shape="1, 16, 1, 1" size="32"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="208" name="bottleneck2_6/dim_red/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck2_6/dim_red/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="209" name="bottleneck2_6/dim_red/fn/weights3111240085" type="Const" version="opset1">
+ <data element_type="f32" offset="2332" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="210" name="bottleneck2_6/dim_red/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck2_6/dim_red/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="211" name="bottleneck2_6/inner/dw1/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f16" offset="39600" shape="16, 1, 1, 3, 3" size="288"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="212" name="bottleneck2_6/inner/dw1/conv" type="GroupConvolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="1, 1" pads_end="1, 1" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="213" name="data_add_2391323918" type="Const" version="opset1">
+ <data element_type="f16" offset="39888" shape="1, 16, 1, 1" size="32"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="214" name="bottleneck2_6/inner/dw1/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck2_6/inner/dw1/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="215" name="bottleneck2_6/inner/dw1/fn/weights3097239791" type="Const" version="opset1">
+ <data element_type="f32" offset="2332" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="216" name="bottleneck2_6/inner/dw1/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck2_6/inner/dw1/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="217" name="bottleneck2_6/dim_inc/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f16" offset="39920" shape="64, 16, 1, 1" size="2048"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>64</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="218" name="bottleneck2_6/dim_inc/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="0, 0" pads_end="0, 0" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>64</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="219" name="data_add_2392123926" type="Const" version="opset1">
+ <data element_type="f16" offset="41968" shape="1, 64, 1, 1" size="128"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="220" name="bottleneck2_6/dim_inc/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck2_6/dim_inc/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="221" name="bottleneck2_6/add" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck2_6/add" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="222" name="bottleneck2_6/fn/weights3100040583" type="Const" version="opset1">
+ <data element_type="f32" offset="2332" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="223" name="bottleneck2_6/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck2_6/add" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="224" name="bottleneck2_7/dim_red/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f16" offset="42096" shape="16, 64, 1, 1" size="2048"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>16</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="225" name="bottleneck2_7/dim_red/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="0, 0" pads_end="0, 0" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>16</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="226" name="data_add_2392923934" type="Const" version="opset1">
+ <data element_type="f16" offset="44144" shape="1, 16, 1, 1" size="32"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="227" name="bottleneck2_7/dim_red/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck2_7/dim_red/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="228" name="bottleneck2_7/dim_red/fn/weights3098039671" type="Const" version="opset1">
+ <data element_type="f32" offset="2332" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="229" name="bottleneck2_7/dim_red/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck2_7/dim_red/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="230" name="bottleneck2_7/inner/dw1/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f16" offset="44176" shape="16, 1, 1, 3, 3" size="288"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="231" name="bottleneck2_7/inner/dw1/conv" type="GroupConvolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="1, 1" pads_end="1, 1" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="232" name="data_add_2393723942" type="Const" version="opset1">
+ <data element_type="f16" offset="44464" shape="1, 16, 1, 1" size="32"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="233" name="bottleneck2_7/inner/dw1/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck2_7/inner/dw1/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="234" name="bottleneck2_7/inner/dw1/fn/weights3083640487" type="Const" version="opset1">
+ <data element_type="f32" offset="2332" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="235" name="bottleneck2_7/inner/dw1/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck2_7/inner/dw1/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="236" name="bottleneck2_7/dim_inc/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f16" offset="44496" shape="64, 16, 1, 1" size="2048"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>64</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="237" name="bottleneck2_7/dim_inc/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="0, 0" pads_end="0, 0" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>64</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="238" name="data_add_2394523950" type="Const" version="opset1">
+ <data element_type="f16" offset="46544" shape="1, 64, 1, 1" size="128"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="239" name="bottleneck2_7/dim_inc/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck2_7/dim_inc/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="240" name="bottleneck2_7/add" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck2_7/add" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="241" name="bottleneck2_7/fn/weights3107639704" type="Const" version="opset1">
+ <data element_type="f32" offset="2332" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="242" name="bottleneck2_7/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck2_7/add" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="243" name="bottleneck2_8/dim_red/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f16" offset="46672" shape="16, 64, 1, 1" size="2048"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>16</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="244" name="bottleneck2_8/dim_red/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="0, 0" pads_end="0, 0" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>16</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="245" name="data_add_2395323958" type="Const" version="opset1">
+ <data element_type="f16" offset="48720" shape="1, 16, 1, 1" size="32"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="246" name="bottleneck2_8/dim_red/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck2_8/dim_red/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="247" name="bottleneck2_8/dim_red/fn/weights3090040466" type="Const" version="opset1">
+ <data element_type="f32" offset="2332" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="248" name="bottleneck2_8/dim_red/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck2_8/dim_red/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="249" name="bottleneck2_8/inner/dw1/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f16" offset="48752" shape="16, 1, 1, 3, 3" size="288"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="250" name="bottleneck2_8/inner/dw1/conv" type="GroupConvolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="1, 1" pads_end="1, 1" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="251" name="data_add_2396123966" type="Const" version="opset1">
+ <data element_type="f16" offset="49040" shape="1, 16, 1, 1" size="32"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="252" name="bottleneck2_8/inner/dw1/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck2_8/inner/dw1/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="253" name="bottleneck2_8/inner/dw1/fn/weights3118040316" type="Const" version="opset1">
+ <data element_type="f32" offset="2332" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="254" name="bottleneck2_8/inner/dw1/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck2_8/inner/dw1/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="255" name="bottleneck2_8/dim_inc/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f16" offset="49072" shape="64, 16, 1, 1" size="2048"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>64</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="256" name="bottleneck2_8/dim_inc/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="0, 0" pads_end="0, 0" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>64</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="257" name="data_add_2396923974" type="Const" version="opset1">
+ <data element_type="f16" offset="51120" shape="1, 64, 1, 1" size="128"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="258" name="bottleneck2_8/dim_inc/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck2_8/dim_inc/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="259" name="bottleneck2_8/add" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck2_8/add" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="260" name="bottleneck2_8/fn/weights3112840481" type="Const" version="opset1">
+ <data element_type="f32" offset="2332" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="261" name="bottleneck2_8/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck2_8/add" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="262" name="bottleneck3_0/skip/pooling" type="MaxPool" version="opset1">
+ <data auto_pad="explicit" kernel="2, 2" pads_begin="0, 0" pads_end="0, 0" rounding_type="ceil" strides="2, 2"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </input>
+ <output>
+ <port id="1" names="bottleneck3_0/skip/pooling" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="263" name="bottleneck3_0/skip/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f16" offset="51248" shape="128, 64, 1, 1" size="16384"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>128</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="264" name="bottleneck3_0/skip/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="0, 0" pads_end="0, 0" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>128</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="265" name="data_add_2397723982" type="Const" version="opset1">
+ <data element_type="f16" offset="67632" shape="1, 128, 1, 1" size="256"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="266" name="bottleneck3_0/skip/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_0/skip/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="267" name="bottleneck3_0/dim_red/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f16" offset="67888" shape="32, 64, 1, 1" size="4096"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>32</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="268" name="bottleneck3_0/dim_red/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="0, 0" pads_end="0, 0" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>32</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="269" name="data_add_2398523990" type="Const" version="opset1">
+ <data element_type="f16" offset="71984" shape="1, 32, 1, 1" size="64"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="270" name="bottleneck3_0/dim_red/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_0/dim_red/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="271" name="bottleneck3_0/dim_red/fn/weights3096440007" type="Const" version="opset1">
+ <data element_type="f32" offset="2332" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="272" name="bottleneck3_0/dim_red/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_0/dim_red/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="273" name="bottleneck3_0/inner/dw1/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f16" offset="72048" shape="32, 1, 1, 3, 3" size="576"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="274" name="bottleneck3_0/inner/dw1/conv" type="GroupConvolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="1, 1" pads_end="1, 1" strides="2, 2"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="275" name="data_add_2399323998" type="Const" version="opset1">
+ <data element_type="f16" offset="72624" shape="1, 32, 1, 1" size="64"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="276" name="bottleneck3_0/inner/dw1/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_0/inner/dw1/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="277" name="bottleneck3_0/inner/dw1/fn/weights3084840604" type="Const" version="opset1">
+ <data element_type="f32" offset="2332" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="278" name="bottleneck3_0/inner/dw1/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_0/inner/dw1/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="279" name="bottleneck3_0/dim_inc/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f16" offset="72688" shape="128, 32, 1, 1" size="8192"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>128</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="280" name="bottleneck3_0/dim_inc/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="0, 0" pads_end="0, 0" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>128</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="281" name="data_add_2400124006" type="Const" version="opset1">
+ <data element_type="f16" offset="80880" shape="1, 128, 1, 1" size="256"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="282" name="bottleneck3_0/dim_inc/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_0/dim_inc/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="283" name="bottleneck3_0/add" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_0/add" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="284" name="bottleneck3_0/fn/weights3096040034" type="Const" version="opset1">
+ <data element_type="f32" offset="2332" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="285" name="bottleneck3_0/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_0/add" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="286" name="bottleneck3_1/dim_red/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f16" offset="81136" shape="32, 128, 1, 1" size="8192"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>32</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="287" name="bottleneck3_1/dim_red/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="0, 0" pads_end="0, 0" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>32</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="288" name="data_add_2400924014" type="Const" version="opset1">
+ <data element_type="f16" offset="89328" shape="1, 32, 1, 1" size="64"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="289" name="bottleneck3_1/dim_red/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_1/dim_red/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="290" name="bottleneck3_1/dim_red/fn/weights3089640577" type="Const" version="opset1">
+ <data element_type="f32" offset="2332" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="291" name="bottleneck3_1/dim_red/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_1/dim_red/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="292" name="bottleneck3_1/inner/dw1/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f16" offset="89392" shape="32, 1, 1, 3, 3" size="576"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="293" name="bottleneck3_1/inner/dw1/conv" type="GroupConvolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="1, 1" pads_end="1, 1" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="294" name="data_add_2401724022" type="Const" version="opset1">
+ <data element_type="f16" offset="89968" shape="1, 32, 1, 1" size="64"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="295" name="bottleneck3_1/inner/dw1/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_1/inner/dw1/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="296" name="bottleneck3_1/inner/dw1/fn/weights3095639668" type="Const" version="opset1">
+ <data element_type="f32" offset="2332" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="297" name="bottleneck3_1/inner/dw1/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_1/inner/dw1/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="298" name="bottleneck3_1/dim_inc/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f16" offset="90032" shape="128, 32, 1, 1" size="8192"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>128</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="299" name="bottleneck3_1/dim_inc/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="0, 0" pads_end="0, 0" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>128</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="300" name="data_add_2402524030" type="Const" version="opset1">
+ <data element_type="f16" offset="98224" shape="1, 128, 1, 1" size="256"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="301" name="bottleneck3_1/dim_inc/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_1/dim_inc/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="302" name="bottleneck3_1/add" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_1/add" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="303" name="bottleneck3_1/fn/weights3088040013" type="Const" version="opset1">
+ <data element_type="f32" offset="2332" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="304" name="bottleneck3_1/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_1/add" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="305" name="bottleneck3_2/dim_red/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f16" offset="98480" shape="32, 128, 1, 1" size="8192"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>32</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="306" name="bottleneck3_2/dim_red/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="0, 0" pads_end="0, 0" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>32</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="307" name="data_add_2403324038" type="Const" version="opset1">
+ <data element_type="f16" offset="106672" shape="1, 32, 1, 1" size="64"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="308" name="bottleneck3_2/dim_red/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_2/dim_red/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="309" name="bottleneck3_2/dim_red/fn/weights3082040079" type="Const" version="opset1">
+ <data element_type="f32" offset="2332" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="310" name="bottleneck3_2/dim_red/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_2/dim_red/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="311" name="bottleneck3_2/inner/dw1/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f16" offset="106736" shape="32, 1, 1, 3, 3" size="576"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="312" name="bottleneck3_2/inner/dw1/conv" type="GroupConvolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="1, 1" pads_end="1, 1" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="313" name="data_add_2404124046" type="Const" version="opset1">
+ <data element_type="f16" offset="107312" shape="1, 32, 1, 1" size="64"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="314" name="bottleneck3_2/inner/dw1/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_2/inner/dw1/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="315" name="bottleneck3_2/inner/dw1/fn/weights3081640064" type="Const" version="opset1">
+ <data element_type="f32" offset="2332" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="316" name="bottleneck3_2/inner/dw1/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_2/inner/dw1/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="317" name="bottleneck3_2/dim_inc/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f16" offset="107376" shape="128, 32, 1, 1" size="8192"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>128</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="318" name="bottleneck3_2/dim_inc/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="0, 0" pads_end="0, 0" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>128</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="319" name="data_add_2404924054" type="Const" version="opset1">
+ <data element_type="f16" offset="115568" shape="1, 128, 1, 1" size="256"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="320" name="bottleneck3_2/dim_inc/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_2/dim_inc/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="321" name="bottleneck3_2/add" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_2/add" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="322" name="bottleneck3_2/fn/weights3088840037" type="Const" version="opset1">
+ <data element_type="f32" offset="2332" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="323" name="bottleneck3_2/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_2/add" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="324" name="bottleneck3_3/dim_red/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f16" offset="115824" shape="32, 128, 1, 1" size="8192"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>32</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="325" name="bottleneck3_3/dim_red/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="0, 0" pads_end="0, 0" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>32</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="326" name="data_add_2405724062" type="Const" version="opset1">
+ <data element_type="f16" offset="124016" shape="1, 32, 1, 1" size="64"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="327" name="bottleneck3_3/dim_red/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_3/dim_red/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="328" name="bottleneck3_3/dim_red/fn/weights3081240370" type="Const" version="opset1">
+ <data element_type="f32" offset="2332" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="329" name="bottleneck3_3/dim_red/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_3/dim_red/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="330" name="bottleneck3_3/inner/dw1/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f16" offset="124080" shape="32, 1, 1, 3, 3" size="576"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="331" name="bottleneck3_3/inner/dw1/conv" type="GroupConvolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="1, 1" pads_end="1, 1" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="332" name="data_add_2406524070" type="Const" version="opset1">
+ <data element_type="f16" offset="124656" shape="1, 32, 1, 1" size="64"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="333" name="bottleneck3_3/inner/dw1/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_3/inner/dw1/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="334" name="bottleneck3_3/inner/dw1/fn/weights3107239839" type="Const" version="opset1">
+ <data element_type="f32" offset="2332" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="335" name="bottleneck3_3/inner/dw1/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_3/inner/dw1/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="336" name="bottleneck3_3/dim_inc/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f16" offset="124720" shape="128, 32, 1, 1" size="8192"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>128</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="337" name="bottleneck3_3/dim_inc/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="0, 0" pads_end="0, 0" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>128</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="338" name="data_add_2407324078" type="Const" version="opset1">
+ <data element_type="f16" offset="132912" shape="1, 128, 1, 1" size="256"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="339" name="bottleneck3_3/dim_inc/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_3/dim_inc/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="340" name="bottleneck3_3/add" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_3/add" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="341" name="bottleneck3_3/fn/weights3096840265" type="Const" version="opset1">
+ <data element_type="f32" offset="2332" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="342" name="bottleneck3_3/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_3/add" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="343" name="bottleneck3_4/dim_red/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f16" offset="133168" shape="32, 128, 1, 1" size="8192"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>32</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="344" name="bottleneck3_4/dim_red/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="0, 0" pads_end="0, 0" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>32</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="345" name="data_add_2408124086" type="Const" version="opset1">
+ <data element_type="f16" offset="141360" shape="1, 32, 1, 1" size="64"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="346" name="bottleneck3_4/dim_red/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_4/dim_red/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="347" name="bottleneck3_4/dim_red/fn/weights3087239794" type="Const" version="opset1">
+ <data element_type="f32" offset="2332" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="348" name="bottleneck3_4/dim_red/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_4/dim_red/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="349" name="bottleneck3_4/inner/dw1/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f16" offset="141424" shape="32, 1, 1, 3, 3" size="576"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="350" name="bottleneck3_4/inner/dw1/conv" type="GroupConvolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="1, 1" pads_end="1, 1" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="351" name="data_add_2408924094" type="Const" version="opset1">
+ <data element_type="f16" offset="142000" shape="1, 32, 1, 1" size="64"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="352" name="bottleneck3_4/inner/dw1/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_4/inner/dw1/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="353" name="bottleneck3_4/inner/dw1/fn/weights3080040640" type="Const" version="opset1">
+ <data element_type="f32" offset="2332" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="354" name="bottleneck3_4/inner/dw1/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_4/inner/dw1/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="355" name="bottleneck3_4/dim_inc/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f16" offset="142064" shape="128, 32, 1, 1" size="8192"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>128</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="356" name="bottleneck3_4/dim_inc/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="0, 0" pads_end="0, 0" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>128</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="357" name="data_add_2409724102" type="Const" version="opset1">
+ <data element_type="f16" offset="150256" shape="1, 128, 1, 1" size="256"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="358" name="bottleneck3_4/dim_inc/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_4/dim_inc/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="359" name="bottleneck3_4/add" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_4/add" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="360" name="bottleneck3_4/fn/weights3089240439" type="Const" version="opset1">
+ <data element_type="f32" offset="2332" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="361" name="bottleneck3_4/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_4/add" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="362" name="bottleneck3_5/dim_red/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f16" offset="150512" shape="32, 128, 1, 1" size="8192"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>32</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="363" name="bottleneck3_5/dim_red/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="0, 0" pads_end="0, 0" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>32</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="364" name="data_add_2410524110" type="Const" version="opset1">
+ <data element_type="f16" offset="158704" shape="1, 32, 1, 1" size="64"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="365" name="bottleneck3_5/dim_red/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_5/dim_red/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="366" name="bottleneck3_5/dim_red/fn/weights3080839863" type="Const" version="opset1">
+ <data element_type="f32" offset="2332" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="367" name="bottleneck3_5/dim_red/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_5/dim_red/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="368" name="bottleneck3_5/inner/dw1/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f16" offset="158768" shape="32, 1, 1, 3, 3" size="576"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="369" name="bottleneck3_5/inner/dw1/conv" type="GroupConvolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="1, 1" pads_end="1, 1" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="370" name="data_add_2411324118" type="Const" version="opset1">
+ <data element_type="f16" offset="159344" shape="1, 32, 1, 1" size="64"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="371" name="bottleneck3_5/inner/dw1/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_5/inner/dw1/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="372" name="bottleneck3_5/inner/dw1/fn/weights3115639692" type="Const" version="opset1">
+ <data element_type="f32" offset="2332" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="373" name="bottleneck3_5/inner/dw1/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_5/inner/dw1/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="374" name="bottleneck3_5/dim_inc/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f16" offset="159408" shape="128, 32, 1, 1" size="8192"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>128</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="375" name="bottleneck3_5/dim_inc/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="0, 0" pads_end="0, 0" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>128</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="376" name="data_add_2412124126" type="Const" version="opset1">
+ <data element_type="f16" offset="167600" shape="1, 128, 1, 1" size="256"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="377" name="bottleneck3_5/dim_inc/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_5/dim_inc/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="378" name="bottleneck3_5/add" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_5/add" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="379" name="bottleneck3_5/fn/weights3084439761" type="Const" version="opset1">
+ <data element_type="f32" offset="2332" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="380" name="bottleneck3_5/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_5/add" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="381" name="bottleneck3_6/dim_red/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f16" offset="167856" shape="32, 128, 1, 1" size="8192"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>32</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="382" name="bottleneck3_6/dim_red/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="0, 0" pads_end="0, 0" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>32</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="383" name="data_add_2412924134" type="Const" version="opset1">
+ <data element_type="f16" offset="176048" shape="1, 32, 1, 1" size="64"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="384" name="bottleneck3_6/dim_red/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_6/dim_red/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="385" name="bottleneck3_6/dim_red/fn/weights3097640670" type="Const" version="opset1">
+ <data element_type="f32" offset="2332" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="386" name="bottleneck3_6/dim_red/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_6/dim_red/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="387" name="bottleneck3_6/inner/dw1/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f16" offset="176112" shape="32, 1, 1, 3, 3" size="576"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="388" name="bottleneck3_6/inner/dw1/conv" type="GroupConvolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="1, 1" pads_end="1, 1" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="389" name="data_add_2413724142" type="Const" version="opset1">
+ <data element_type="f16" offset="176688" shape="1, 32, 1, 1" size="64"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="390" name="bottleneck3_6/inner/dw1/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_6/inner/dw1/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="391" name="bottleneck3_6/inner/dw1/fn/weights3114040469" type="Const" version="opset1">
+ <data element_type="f32" offset="2332" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="392" name="bottleneck3_6/inner/dw1/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_6/inner/dw1/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="393" name="bottleneck3_6/dim_inc/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f16" offset="176752" shape="128, 32, 1, 1" size="8192"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>128</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="394" name="bottleneck3_6/dim_inc/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="0, 0" pads_end="0, 0" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>128</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="395" name="data_add_2414524150" type="Const" version="opset1">
+ <data element_type="f16" offset="184944" shape="1, 128, 1, 1" size="256"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="396" name="bottleneck3_6/dim_inc/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_6/dim_inc/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="397" name="bottleneck3_6/add" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_6/add" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="398" name="bottleneck3_6/fn/weights3099239857" type="Const" version="opset1">
+ <data element_type="f32" offset="2332" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="399" name="bottleneck3_6/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_6/add" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="400" name="bottleneck3_7/dim_red/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f16" offset="185200" shape="32, 128, 1, 1" size="8192"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>32</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="401" name="bottleneck3_7/dim_red/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="0, 0" pads_end="0, 0" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>32</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="402" name="data_add_2415324158" type="Const" version="opset1">
+ <data element_type="f16" offset="193392" shape="1, 32, 1, 1" size="64"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="403" name="bottleneck3_7/dim_red/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_7/dim_red/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="404" name="bottleneck3_7/dim_red/fn/weights3106039974" type="Const" version="opset1">
+ <data element_type="f32" offset="2332" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="405" name="bottleneck3_7/dim_red/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_7/dim_red/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="406" name="bottleneck3_7/inner/dw1/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f16" offset="193456" shape="32, 1, 1, 3, 3" size="576"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="407" name="bottleneck3_7/inner/dw1/conv" type="GroupConvolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="1, 1" pads_end="1, 1" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="408" name="data_add_2416124166" type="Const" version="opset1">
+ <data element_type="f16" offset="194032" shape="1, 32, 1, 1" size="64"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="409" name="bottleneck3_7/inner/dw1/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_7/inner/dw1/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="410" name="bottleneck3_7/inner/dw1/fn/weights3104039701" type="Const" version="opset1">
+ <data element_type="f32" offset="2332" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="411" name="bottleneck3_7/inner/dw1/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_7/inner/dw1/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="412" name="bottleneck3_7/dim_inc/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f16" offset="194096" shape="128, 32, 1, 1" size="8192"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>128</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="413" name="bottleneck3_7/dim_inc/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="0, 0" pads_end="0, 0" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>128</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="414" name="data_add_2416924174" type="Const" version="opset1">
+ <data element_type="f16" offset="202288" shape="1, 128, 1, 1" size="256"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="415" name="bottleneck3_7/dim_inc/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_7/dim_inc/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="416" name="bottleneck3_7/add" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_7/add" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="417" name="bottleneck3_7/fn/weights3079640682" type="Const" version="opset1">
+ <data element_type="f32" offset="2332" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="418" name="bottleneck3_7/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_7/add" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="419" name="bottleneck3_8/dim_red/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f16" offset="202544" shape="32, 128, 1, 1" size="8192"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>32</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="420" name="bottleneck3_8/dim_red/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="0, 0" pads_end="0, 0" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>32</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="421" name="data_add_2417724182" type="Const" version="opset1">
+ <data element_type="f16" offset="210736" shape="1, 32, 1, 1" size="64"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="422" name="bottleneck3_8/dim_red/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_8/dim_red/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="423" name="bottleneck3_8/dim_red/fn/weights3112439755" type="Const" version="opset1">
+ <data element_type="f32" offset="2332" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="424" name="bottleneck3_8/dim_red/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_8/dim_red/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="425" name="bottleneck3_8/inner/dw1/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f16" offset="210800" shape="32, 1, 1, 3, 3" size="576"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="426" name="bottleneck3_8/inner/dw1/conv" type="GroupConvolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="1, 1" pads_end="1, 1" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="427" name="data_add_2418524190" type="Const" version="opset1">
+ <data element_type="f16" offset="211376" shape="1, 32, 1, 1" size="64"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="428" name="bottleneck3_8/inner/dw1/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_8/inner/dw1/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="429" name="bottleneck3_8/inner/dw1/fn/weights3092839893" type="Const" version="opset1">
+ <data element_type="f32" offset="2332" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="430" name="bottleneck3_8/inner/dw1/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_8/inner/dw1/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="431" name="bottleneck3_8/dim_inc/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f16" offset="211440" shape="128, 32, 1, 1" size="8192"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>128</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="432" name="bottleneck3_8/dim_inc/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="0, 0" pads_end="0, 0" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>128</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="433" name="data_add_2419324198" type="Const" version="opset1">
+ <data element_type="f16" offset="219632" shape="1, 128, 1, 1" size="256"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="434" name="bottleneck3_8/dim_inc/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_8/dim_inc/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="435" name="bottleneck3_8/add" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_8/add" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="436" name="bottleneck3_8/fn/weights3109640649" type="Const" version="opset1">
+ <data element_type="f32" offset="2332" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="437" name="bottleneck3_8/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_8/add" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="438" name="bottleneck3_9/dim_red/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f16" offset="219888" shape="32, 128, 1, 1" size="8192"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>32</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="439" name="bottleneck3_9/dim_red/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="0, 0" pads_end="0, 0" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>32</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="440" name="data_add_2420124206" type="Const" version="opset1">
+ <data element_type="f16" offset="228080" shape="1, 32, 1, 1" size="64"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="441" name="bottleneck3_9/dim_red/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_9/dim_red/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="442" name="bottleneck3_9/dim_red/fn/weights3083240112" type="Const" version="opset1">
+ <data element_type="f32" offset="2332" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="443" name="bottleneck3_9/dim_red/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_9/dim_red/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="444" name="bottleneck3_9/inner/dw1/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f16" offset="228144" shape="32, 1, 1, 3, 3" size="576"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="445" name="bottleneck3_9/inner/dw1/conv" type="GroupConvolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="1, 1" pads_end="1, 1" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="446" name="data_add_2420924214" type="Const" version="opset1">
+ <data element_type="f16" offset="228720" shape="1, 32, 1, 1" size="64"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="447" name="bottleneck3_9/inner/dw1/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_9/inner/dw1/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="448" name="bottleneck3_9/inner/dw1/fn/weights3104840376" type="Const" version="opset1">
+ <data element_type="f32" offset="2332" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="449" name="bottleneck3_9/inner/dw1/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_9/inner/dw1/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="450" name="bottleneck3_9/dim_inc/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f16" offset="228784" shape="128, 32, 1, 1" size="8192"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>128</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="451" name="bottleneck3_9/dim_inc/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="0, 0" pads_end="0, 0" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>128</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="452" name="data_add_2421724222" type="Const" version="opset1">
+ <data element_type="f16" offset="236976" shape="1, 128, 1, 1" size="256"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="453" name="bottleneck3_9/dim_inc/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_9/dim_inc/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="454" name="bottleneck3_9/add" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_9/add" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="455" name="bottleneck3_9/fn/weights3094040340" type="Const" version="opset1">
+ <data element_type="f32" offset="2332" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="456" name="bottleneck3_9/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_9/add" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="457" name="bottleneck3_10/dim_red/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f16" offset="237232" shape="32, 128, 1, 1" size="8192"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>32</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="458" name="bottleneck3_10/dim_red/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="0, 0" pads_end="0, 0" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>32</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="459" name="data_add_2422524230" type="Const" version="opset1">
+ <data element_type="f16" offset="245424" shape="1, 32, 1, 1" size="64"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="460" name="bottleneck3_10/dim_red/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_10/dim_red/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="461" name="bottleneck3_10/dim_red/fn/weights3082839743" type="Const" version="opset1">
+ <data element_type="f32" offset="2332" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="462" name="bottleneck3_10/dim_red/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_10/dim_red/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="463" name="bottleneck3_10/inner/dw1/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f16" offset="245488" shape="32, 1, 1, 3, 3" size="576"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="464" name="bottleneck3_10/inner/dw1/conv" type="GroupConvolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="1, 1" pads_end="1, 1" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="465" name="data_add_2423324238" type="Const" version="opset1">
+ <data element_type="f16" offset="246064" shape="1, 32, 1, 1" size="64"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="466" name="bottleneck3_10/inner/dw1/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_10/inner/dw1/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="467" name="bottleneck3_10/inner/dw1/fn/weights3099639836" type="Const" version="opset1">
+ <data element_type="f32" offset="2332" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="468" name="bottleneck3_10/inner/dw1/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_10/inner/dw1/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="469" name="bottleneck3_10/dim_inc/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f16" offset="246128" shape="128, 32, 1, 1" size="8192"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>128</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="470" name="bottleneck3_10/dim_inc/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="0, 0" pads_end="0, 0" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>128</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="471" name="data_add_2424124246" type="Const" version="opset1">
+ <data element_type="f16" offset="254320" shape="1, 128, 1, 1" size="256"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="472" name="bottleneck3_10/dim_inc/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_10/dim_inc/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="473" name="bottleneck3_10/add" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_10/add" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="474" name="bottleneck3_10/fn/weights3077239977" type="Const" version="opset1">
+ <data element_type="f32" offset="2332" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="475" name="bottleneck3_10/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_10/add" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="476" name="bottleneck4_0/skip/pooling" type="MaxPool" version="opset1">
+ <data auto_pad="explicit" kernel="2, 2" pads_begin="0, 0" pads_end="0, 0" rounding_type="ceil" strides="2, 2"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </input>
+ <output>
+ <port id="1" names="bottleneck4_0/skip/pooling" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="477" name="bottleneck4_0/skip/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f16" offset="254576" shape="256, 128, 1, 1" size="65536"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>256</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="478" name="bottleneck4_0/skip/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="0, 0" pads_end="0, 0" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>256</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="479" name="data_add_2424924254" type="Const" version="opset1">
+ <data element_type="f16" offset="320112" shape="1, 256, 1, 1" size="512"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="480" name="bottleneck4_0/skip/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_0/skip/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="481" name="bottleneck4_0/dim_red/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f16" offset="320624" shape="64, 128, 1, 1" size="16384"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>64</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="482" name="bottleneck4_0/dim_red/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="0, 0" pads_end="0, 0" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>64</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="483" name="data_add_2425724262" type="Const" version="opset1">
+ <data element_type="f16" offset="337008" shape="1, 64, 1, 1" size="128"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="484" name="bottleneck4_0/dim_red/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_0/dim_red/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="485" name="bottleneck4_0/dim_red/fn/weights3101639809" type="Const" version="opset1">
+ <data element_type="f32" offset="2332" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="486" name="bottleneck4_0/dim_red/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_0/dim_red/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="487" name="bottleneck4_0/inner/dw1/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f16" offset="337136" shape="64, 1, 1, 3, 3" size="1152"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="488" name="bottleneck4_0/inner/dw1/conv" type="GroupConvolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="1, 1" pads_end="1, 1" strides="2, 2"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="489" name="data_add_2426524270" type="Const" version="opset1">
+ <data element_type="f16" offset="338288" shape="1, 64, 1, 1" size="128"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="490" name="bottleneck4_0/inner/dw1/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_0/inner/dw1/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="491" name="bottleneck4_0/inner/dw1/fn/weights3082440412" type="Const" version="opset1">
+ <data element_type="f32" offset="2332" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="492" name="bottleneck4_0/inner/dw1/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_0/inner/dw1/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="493" name="bottleneck4_0/dim_inc/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f16" offset="338416" shape="256, 64, 1, 1" size="32768"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>256</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="494" name="bottleneck4_0/dim_inc/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="0, 0" pads_end="0, 0" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>256</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="495" name="data_add_2427324278" type="Const" version="opset1">
+ <data element_type="f16" offset="371184" shape="1, 256, 1, 1" size="512"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="496" name="bottleneck4_0/dim_inc/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_0/dim_inc/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="497" name="bottleneck4_0/add" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_0/add" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="498" name="bottleneck4_0/fn/weights3103639980" type="Const" version="opset1">
+ <data element_type="f32" offset="2332" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="499" name="bottleneck4_0/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_0/add" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="500" name="bottleneck4_1/dim_red/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f16" offset="371696" shape="64, 256, 1, 1" size="32768"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>64</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="501" name="bottleneck4_1/dim_red/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="0, 0" pads_end="0, 0" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>64</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="502" name="data_add_2428124286" type="Const" version="opset1">
+ <data element_type="f16" offset="404464" shape="1, 64, 1, 1" size="128"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="503" name="bottleneck4_1/dim_red/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_1/dim_red/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="504" name="bottleneck4_1/dim_red/fn/weights3100440331" type="Const" version="opset1">
+ <data element_type="f32" offset="2332" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="505" name="bottleneck4_1/dim_red/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_1/dim_red/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="506" name="bottleneck4_1/inner/dw1/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f16" offset="404592" shape="64, 1, 1, 3, 3" size="1152"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="507" name="bottleneck4_1/inner/dw1/conv" type="GroupConvolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="1, 1" pads_end="1, 1" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="508" name="data_add_2428924294" type="Const" version="opset1">
+ <data element_type="f16" offset="405744" shape="1, 64, 1, 1" size="128"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="509" name="bottleneck4_1/inner/dw1/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_1/inner/dw1/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="510" name="bottleneck4_1/inner/dw1/fn/weights3108840358" type="Const" version="opset1">
+ <data element_type="f32" offset="2332" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="511" name="bottleneck4_1/inner/dw1/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_1/inner/dw1/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="512" name="bottleneck4_1/dim_inc/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f16" offset="405872" shape="256, 64, 1, 1" size="32768"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>256</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="513" name="bottleneck4_1/dim_inc/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="0, 0" pads_end="0, 0" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>256</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="514" name="data_add_2429724302" type="Const" version="opset1">
+ <data element_type="f16" offset="438640" shape="1, 256, 1, 1" size="512"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="515" name="bottleneck4_1/dim_inc/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_1/dim_inc/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="516" name="bottleneck4_1/add" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_1/add" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="517" name="bottleneck4_1/fn/weights3114439767" type="Const" version="opset1">
+ <data element_type="f32" offset="2332" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="518" name="bottleneck4_1/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_1/add" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="519" name="bottleneck4_2/dim_red/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f16" offset="439152" shape="64, 256, 1, 1" size="32768"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>64</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="520" name="bottleneck4_2/dim_red/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="0, 0" pads_end="0, 0" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>64</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="521" name="data_add_2430524310" type="Const" version="opset1">
+ <data element_type="f16" offset="471920" shape="1, 64, 1, 1" size="128"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="522" name="bottleneck4_2/dim_red/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_2/dim_red/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="523" name="bottleneck4_2/dim_red/fn/weights3076039947" type="Const" version="opset1">
+ <data element_type="f32" offset="2332" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="524" name="bottleneck4_2/dim_red/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_2/dim_red/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="525" name="bottleneck4_2/inner/dw1/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f16" offset="472048" shape="64, 1, 1, 3, 3" size="1152"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="526" name="bottleneck4_2/inner/dw1/conv" type="GroupConvolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="1, 1" pads_end="1, 1" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="527" name="data_add_2431324318" type="Const" version="opset1">
+ <data element_type="f16" offset="473200" shape="1, 64, 1, 1" size="128"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="528" name="bottleneck4_2/inner/dw1/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_2/inner/dw1/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="529" name="bottleneck4_2/inner/dw1/fn/weights3085640283" type="Const" version="opset1">
+ <data element_type="f32" offset="2332" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="530" name="bottleneck4_2/inner/dw1/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_2/inner/dw1/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="531" name="bottleneck4_2/dim_inc/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f16" offset="473328" shape="256, 64, 1, 1" size="32768"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>256</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="532" name="bottleneck4_2/dim_inc/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="0, 0" pads_end="0, 0" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>256</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="533" name="data_add_2432124326" type="Const" version="opset1">
+ <data element_type="f16" offset="506096" shape="1, 256, 1, 1" size="512"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="534" name="bottleneck4_2/dim_inc/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_2/dim_inc/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="535" name="bottleneck4_2/add" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_2/add" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="536" name="bottleneck4_2/fn/weights3108440136" type="Const" version="opset1">
+ <data element_type="f32" offset="2332" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="537" name="bottleneck4_2/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_2/add" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="538" name="bottleneck4_3/dim_red/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f16" offset="506608" shape="64, 256, 1, 1" size="32768"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>64</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="539" name="bottleneck4_3/dim_red/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="0, 0" pads_end="0, 0" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>64</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="540" name="data_add_2432924334" type="Const" version="opset1">
+ <data element_type="f16" offset="539376" shape="1, 64, 1, 1" size="128"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="541" name="bottleneck4_3/dim_red/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_3/dim_red/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="542" name="bottleneck4_3/dim_red/fn/weights3079239749" type="Const" version="opset1">
+ <data element_type="f32" offset="2332" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="543" name="bottleneck4_3/dim_red/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_3/dim_red/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="544" name="bottleneck4_3/inner/dw1/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f16" offset="539504" shape="64, 1, 1, 3, 3" size="1152"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="545" name="bottleneck4_3/inner/dw1/conv" type="GroupConvolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="1, 1" pads_end="1, 1" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="546" name="data_add_2433724342" type="Const" version="opset1">
+ <data element_type="f16" offset="540656" shape="1, 64, 1, 1" size="128"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="547" name="bottleneck4_3/inner/dw1/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_3/inner/dw1/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="548" name="bottleneck4_3/inner/dw1/fn/weights3095239962" type="Const" version="opset1">
+ <data element_type="f32" offset="2332" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="549" name="bottleneck4_3/inner/dw1/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_3/inner/dw1/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="550" name="bottleneck4_3/dim_inc/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f16" offset="540784" shape="256, 64, 1, 1" size="32768"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>256</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="551" name="bottleneck4_3/dim_inc/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="0, 0" pads_end="0, 0" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>256</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="552" name="data_add_2434524350" type="Const" version="opset1">
+ <data element_type="f16" offset="573552" shape="1, 256, 1, 1" size="512"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="553" name="bottleneck4_3/dim_inc/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_3/dim_inc/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="554" name="bottleneck4_3/add" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_3/add" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="555" name="bottleneck4_3/fn/weights3078440184" type="Const" version="opset1">
+ <data element_type="f32" offset="2332" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="556" name="bottleneck4_3/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_3/add" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="557" name="bottleneck4_4/dim_red/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f16" offset="574064" shape="64, 256, 1, 1" size="32768"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>64</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="558" name="bottleneck4_4/dim_red/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="0, 0" pads_end="0, 0" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>64</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="559" name="data_add_2435324358" type="Const" version="opset1">
+ <data element_type="f16" offset="606832" shape="1, 64, 1, 1" size="128"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="560" name="bottleneck4_4/dim_red/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_4/dim_red/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="561" name="bottleneck4_4/dim_red/fn/weights3093240379" type="Const" version="opset1">
+ <data element_type="f32" offset="2332" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="562" name="bottleneck4_4/dim_red/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_4/dim_red/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="563" name="bottleneck4_4/inner/dw1/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f16" offset="606960" shape="64, 1, 1, 3, 3" size="1152"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="564" name="bottleneck4_4/inner/dw1/conv" type="GroupConvolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="1, 1" pads_end="1, 1" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="565" name="data_add_2436124366" type="Const" version="opset1">
+ <data element_type="f16" offset="608112" shape="1, 64, 1, 1" size="128"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="566" name="bottleneck4_4/inner/dw1/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_4/inner/dw1/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="567" name="bottleneck4_4/inner/dw1/fn/weights3116440067" type="Const" version="opset1">
+ <data element_type="f32" offset="2332" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="568" name="bottleneck4_4/inner/dw1/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_4/inner/dw1/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="569" name="bottleneck4_4/dim_inc/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f16" offset="608240" shape="256, 64, 1, 1" size="32768"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>256</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="570" name="bottleneck4_4/dim_inc/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="0, 0" pads_end="0, 0" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>256</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="571" name="data_add_2436924374" type="Const" version="opset1">
+ <data element_type="f16" offset="641008" shape="1, 256, 1, 1" size="512"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="572" name="bottleneck4_4/dim_inc/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_4/dim_inc/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="573" name="bottleneck4_4/add" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_4/add" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="574" name="bottleneck4_4/fn/weights3090840436" type="Const" version="opset1">
+ <data element_type="f32" offset="2332" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="575" name="bottleneck4_4/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_4/add" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="576" name="bottleneck4_5/dim_red/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f16" offset="641520" shape="64, 256, 1, 1" size="32768"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>64</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="577" name="bottleneck4_5/dim_red/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="0, 0" pads_end="0, 0" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>64</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="578" name="data_add_2437724382" type="Const" version="opset1">
+ <data element_type="f16" offset="674288" shape="1, 64, 1, 1" size="128"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="579" name="bottleneck4_5/dim_red/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_5/dim_red/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="580" name="bottleneck4_5/dim_red/fn/weights3078840361" type="Const" version="opset1">
+ <data element_type="f32" offset="2332" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="581" name="bottleneck4_5/dim_red/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_5/dim_red/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="582" name="bottleneck4_5/inner/dw1/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f16" offset="674416" shape="64, 1, 1, 3, 3" size="1152"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="583" name="bottleneck4_5/inner/dw1/conv" type="GroupConvolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="1, 1" pads_end="1, 1" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="584" name="data_add_2438524390" type="Const" version="opset1">
+ <data element_type="f16" offset="675568" shape="1, 64, 1, 1" size="128"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="585" name="bottleneck4_5/inner/dw1/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_5/inner/dw1/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="586" name="bottleneck4_5/inner/dw1/fn/weights3088439752" type="Const" version="opset1">
+ <data element_type="f32" offset="2332" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="587" name="bottleneck4_5/inner/dw1/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_5/inner/dw1/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="588" name="bottleneck4_5/dim_inc/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f16" offset="675696" shape="256, 64, 1, 1" size="32768"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>256</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="589" name="bottleneck4_5/dim_inc/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="0, 0" pads_end="0, 0" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>256</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="590" name="data_add_2439324398" type="Const" version="opset1">
+ <data element_type="f16" offset="708464" shape="1, 256, 1, 1" size="512"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="591" name="bottleneck4_5/dim_inc/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_5/dim_inc/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="592" name="bottleneck4_5/add" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_5/add" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="593" name="bottleneck4_5/fn/weights3104440607" type="Const" version="opset1">
+ <data element_type="f32" offset="2332" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="594" name="bottleneck4_5/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_5/add" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="595" name="bottleneck4_6/dim_red/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f16" offset="708976" shape="64, 256, 1, 1" size="32768"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>64</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="596" name="bottleneck4_6/dim_red/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="0, 0" pads_end="0, 0" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>64</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="597" name="data_add_2440124406" type="Const" version="opset1">
+ <data element_type="f16" offset="741744" shape="1, 64, 1, 1" size="128"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="598" name="bottleneck4_6/dim_red/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_6/dim_red/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="599" name="bottleneck4_6/dim_red/fn/weights3087640250" type="Const" version="opset1">
+ <data element_type="f32" offset="2332" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="600" name="bottleneck4_6/dim_red/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_6/dim_red/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="601" name="bottleneck4_6/inner/dw1/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f16" offset="741872" shape="64, 1, 1, 3, 3" size="1152"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="602" name="bottleneck4_6/inner/dw1/conv" type="GroupConvolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="1, 1" pads_end="1, 1" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="603" name="data_add_2440924414" type="Const" version="opset1">
+ <data element_type="f16" offset="743024" shape="1, 64, 1, 1" size="128"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="604" name="bottleneck4_6/inner/dw1/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_6/inner/dw1/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="605" name="bottleneck4_6/inner/dw1/fn/weights3092439872" type="Const" version="opset1">
+ <data element_type="f32" offset="2332" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="606" name="bottleneck4_6/inner/dw1/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_6/inner/dw1/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="607" name="bottleneck4_6/dim_inc/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f16" offset="743152" shape="256, 64, 1, 1" size="32768"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>256</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="608" name="bottleneck4_6/dim_inc/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="0, 0" pads_end="0, 0" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>256</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="609" name="data_add_2441724422" type="Const" version="opset1">
+ <data element_type="f16" offset="775920" shape="1, 256, 1, 1" size="512"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="610" name="bottleneck4_6/dim_inc/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_6/dim_inc/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="611" name="bottleneck4_6/add" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_6/add" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="612" name="bottleneck4_6/fn/weights3086840208" type="Const" version="opset1">
+ <data element_type="f32" offset="2332" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="613" name="bottleneck4_6/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_6/add" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="614" name="bottleneck4_7/dim_red/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f16" offset="776432" shape="64, 256, 1, 1" size="32768"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>64</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="615" name="bottleneck4_7/dim_red/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="0, 0" pads_end="0, 0" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>64</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="616" name="data_add_2442524430" type="Const" version="opset1">
+ <data element_type="f16" offset="809200" shape="1, 64, 1, 1" size="128"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="617" name="bottleneck4_7/dim_red/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_7/dim_red/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="618" name="bottleneck4_7/dim_red/fn/weights3106440391" type="Const" version="opset1">
+ <data element_type="f32" offset="2332" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="619" name="bottleneck4_7/dim_red/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_7/dim_red/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="620" name="bottleneck4_7/inner/dw1/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f16" offset="809328" shape="64, 1, 1, 3, 3" size="1152"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="621" name="bottleneck4_7/inner/dw1/conv" type="GroupConvolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="1, 1" pads_end="1, 1" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="622" name="data_add_2443324438" type="Const" version="opset1">
+ <data element_type="f16" offset="810480" shape="1, 64, 1, 1" size="128"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="623" name="bottleneck4_7/inner/dw1/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_7/inner/dw1/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="624" name="bottleneck4_7/inner/dw1/fn/weights3086039971" type="Const" version="opset1">
+ <data element_type="f32" offset="2332" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="625" name="bottleneck4_7/inner/dw1/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_7/inner/dw1/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="626" name="bottleneck4_7/dim_inc/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f16" offset="810608" shape="256, 64, 1, 1" size="32768"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>256</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="627" name="bottleneck4_7/dim_inc/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="0, 0" pads_end="0, 0" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>256</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="628" name="data_add_2444124446" type="Const" version="opset1">
+ <data element_type="f16" offset="843376" shape="1, 256, 1, 1" size="512"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="629" name="bottleneck4_7/dim_inc/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_7/dim_inc/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="630" name="bottleneck4_7/add" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_7/add" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="631" name="bottleneck4_7/fn/weights3085240700" type="Const" version="opset1">
+ <data element_type="f32" offset="2332" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="632" name="bottleneck4_7/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_7/add" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="633" name="bottleneck4_8/dim_red/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f16" offset="843888" shape="64, 256, 1, 1" size="32768"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>64</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="634" name="bottleneck4_8/dim_red/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="0, 0" pads_end="0, 0" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>64</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="635" name="data_add_2444924454" type="Const" version="opset1">
+ <data element_type="f16" offset="876656" shape="1, 64, 1, 1" size="128"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="636" name="bottleneck4_8/dim_red/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_8/dim_red/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="637" name="bottleneck4_8/dim_red/fn/weights3077640592" type="Const" version="opset1">
+ <data element_type="f32" offset="2332" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="638" name="bottleneck4_8/dim_red/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_8/dim_red/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="639" name="bottleneck4_8/inner/dw1/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f16" offset="876784" shape="64, 1, 1, 3, 3" size="1152"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="640" name="bottleneck4_8/inner/dw1/conv" type="GroupConvolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="1, 1" pads_end="1, 1" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="641" name="data_add_2445724462" type="Const" version="opset1">
+ <data element_type="f16" offset="877936" shape="1, 64, 1, 1" size="128"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="642" name="bottleneck4_8/inner/dw1/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_8/inner/dw1/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="643" name="bottleneck4_8/inner/dw1/fn/weights3102440382" type="Const" version="opset1">
+ <data element_type="f32" offset="2332" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="644" name="bottleneck4_8/inner/dw1/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_8/inner/dw1/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="645" name="bottleneck4_8/dim_inc/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f16" offset="878064" shape="256, 64, 1, 1" size="32768"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>256</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="646" name="bottleneck4_8/dim_inc/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="0, 0" pads_end="0, 0" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>256</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="647" name="data_add_2446524470" type="Const" version="opset1">
+ <data element_type="f16" offset="910832" shape="1, 256, 1, 1" size="512"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="648" name="bottleneck4_8/dim_inc/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_8/dim_inc/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="649" name="bottleneck4_8/add" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_8/add" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="650" name="bottleneck4_8/fn/weights3078040133" type="Const" version="opset1">
+ <data element_type="f32" offset="2332" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="651" name="bottleneck4_8/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_8/add" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="652" name="bottleneck4_9/dim_red/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f16" offset="911344" shape="64, 256, 1, 1" size="32768"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>64</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="653" name="bottleneck4_9/dim_red/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="0, 0" pads_end="0, 0" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>64</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="654" name="data_add_2447324478" type="Const" version="opset1">
+ <data element_type="f16" offset="944112" shape="1, 64, 1, 1" size="128"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="655" name="bottleneck4_9/dim_red/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_9/dim_red/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="656" name="bottleneck4_9/dim_red/fn/weights3116040616" type="Const" version="opset1">
+ <data element_type="f32" offset="2332" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="657" name="bottleneck4_9/dim_red/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_9/dim_red/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="658" name="bottleneck4_9/inner/dw1/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f16" offset="944240" shape="64, 1, 1, 3, 3" size="1152"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="659" name="bottleneck4_9/inner/dw1/conv" type="GroupConvolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="1, 1" pads_end="1, 1" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="660" name="data_add_2448124486" type="Const" version="opset1">
+ <data element_type="f16" offset="945392" shape="1, 64, 1, 1" size="128"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="661" name="bottleneck4_9/inner/dw1/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_9/inner/dw1/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="662" name="bottleneck4_9/inner/dw1/fn/weights3084040256" type="Const" version="opset1">
+ <data element_type="f32" offset="2332" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="663" name="bottleneck4_9/inner/dw1/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_9/inner/dw1/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="664" name="bottleneck4_9/dim_inc/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f16" offset="945520" shape="256, 64, 1, 1" size="32768"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>256</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="665" name="bottleneck4_9/dim_inc/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="0, 0" pads_end="0, 0" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>256</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="666" name="data_add_2448924494" type="Const" version="opset1">
+ <data element_type="f16" offset="978288" shape="1, 256, 1, 1" size="512"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="667" name="bottleneck4_9/dim_inc/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_9/dim_inc/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="668" name="bottleneck4_9/add" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_9/add" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="669" name="bottleneck4_9/fn/weights3109239740" type="Const" version="opset1">
+ <data element_type="f32" offset="2332" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="670" name="bottleneck4_9/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_9/add" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="671" name="bottleneck4_10/dim_red/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f16" offset="978800" shape="64, 256, 1, 1" size="32768"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>64</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="672" name="bottleneck4_10/dim_red/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="0, 0" pads_end="0, 0" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>64</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="673" name="data_add_2449724502" type="Const" version="opset1">
+ <data element_type="f16" offset="1011568" shape="1, 64, 1, 1" size="128"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="674" name="bottleneck4_10/dim_red/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_10/dim_red/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="675" name="bottleneck4_10/dim_red/fn/weights3118440457" type="Const" version="opset1">
+ <data element_type="f32" offset="2332" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="676" name="bottleneck4_10/dim_red/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_10/dim_red/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="677" name="bottleneck4_10/inner/dw1/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f16" offset="1011696" shape="64, 1, 1, 3, 3" size="1152"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="678" name="bottleneck4_10/inner/dw1/conv" type="GroupConvolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="1, 1" pads_end="1, 1" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="679" name="data_add_2450524510" type="Const" version="opset1">
+ <data element_type="f16" offset="1012848" shape="1, 64, 1, 1" size="128"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="680" name="bottleneck4_10/inner/dw1/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_10/inner/dw1/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="681" name="bottleneck4_10/inner/dw1/fn/weights3086440388" type="Const" version="opset1">
+ <data element_type="f32" offset="2332" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="682" name="bottleneck4_10/inner/dw1/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_10/inner/dw1/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="683" name="bottleneck4_10/dim_inc/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f16" offset="1012976" shape="256, 64, 1, 1" size="32768"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>256</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="684" name="bottleneck4_10/dim_inc/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="0, 0" pads_end="0, 0" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>256</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="685" name="data_add_2451324518" type="Const" version="opset1">
+ <data element_type="f16" offset="1045744" shape="1, 256, 1, 1" size="512"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="686" name="bottleneck4_10/dim_inc/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_10/dim_inc/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="687" name="bottleneck4_10/add" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_10/add" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="688" name="bottleneck4_10/fn/weights3094840226" type="Const" version="opset1">
+ <data element_type="f32" offset="2332" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="689" name="bottleneck4_10/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_10/add" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="690" name="bottleneck4_11/dim_red/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f16" offset="1046256" shape="64, 256, 1, 1" size="32768"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>64</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="691" name="bottleneck4_11/dim_red/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="0, 0" pads_end="0, 0" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>64</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="692" name="data_add_2452124526" type="Const" version="opset1">
+ <data element_type="f16" offset="1079024" shape="1, 64, 1, 1" size="128"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="693" name="bottleneck4_11/dim_red/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_11/dim_red/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="694" name="bottleneck4_11/dim_red/fn/weights3118840166" type="Const" version="opset1">
+ <data element_type="f32" offset="2332" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="695" name="bottleneck4_11/dim_red/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_11/dim_red/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="696" name="bottleneck4_11/inner/dw1/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f16" offset="1079152" shape="64, 1, 1, 3, 3" size="1152"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="697" name="bottleneck4_11/inner/dw1/conv" type="GroupConvolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="1, 1" pads_end="1, 1" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="698" name="data_add_2452924534" type="Const" version="opset1">
+ <data element_type="f16" offset="1080304" shape="1, 64, 1, 1" size="128"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="699" name="bottleneck4_11/inner/dw1/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_11/inner/dw1/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="700" name="bottleneck4_11/inner/dw1/fn/weights3093640247" type="Const" version="opset1">
+ <data element_type="f32" offset="2332" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="701" name="bottleneck4_11/inner/dw1/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_11/inner/dw1/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="702" name="bottleneck4_11/dim_inc/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f16" offset="1080432" shape="256, 64, 1, 1" size="32768"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>256</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="703" name="bottleneck4_11/dim_inc/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="0, 0" pads_end="0, 0" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>256</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="704" name="data_add_2453724542" type="Const" version="opset1">
+ <data element_type="f16" offset="1113200" shape="1, 256, 1, 1" size="512"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="705" name="bottleneck4_11/dim_inc/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_11/dim_inc/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="706" name="bottleneck4_11/add" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_11/add" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="707" name="bottleneck4_11/fn/weights3110840010" type="Const" version="opset1">
+ <data element_type="f32" offset="2332" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="708" name="bottleneck4_11/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bb_16xout_pd" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="709" name="1166" type="Const" version="opset1">
+ <data element_type="f16" offset="1113712" shape="48, 256, 3, 3" size="221184"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>48</dim>
+ <dim>256</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="710" name="mbox_loc1/out/conv/WithoutBiases" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="1, 1" pads_end="1, 1" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>48</dim>
+ <dim>256</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>48</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="711" name="mbox_loc1/out/conv/Dims13825" type="Const" version="opset1">
+ <data element_type="f16" offset="1334896" shape="1, 48, 1, 1" size="96"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>48</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="712" name="mbox_loc1/out/conv" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>48</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>1</dim>
+ <dim>48</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="mbox_loc1/out/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>48</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="713" name="1296" type="Const" version="opset1">
+ <data element_type="i64" offset="1334992" shape="4" size="32"/>
+ <output>
+ <port id="0" precision="I64">
+ <dim>4</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="714" name="mbox_loc1/out/conv/perm" type="Transpose" version="opset1">
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>48</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="I64">
+ <dim>4</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="mbox_loc1/out/conv/perm" precision="FP16">
+ <dim>1</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ <dim>48</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="715" name="1308/shapes_concat" type="Const" version="opset1">
+ <data element_type="i64" offset="1335024" shape="2" size="16"/>
+ <output>
+ <port id="0" precision="I64">
+ <dim>2</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="716" name="mbox_loc1/out/conv/flat" type="Reshape" version="opset1">
+ <data special_zero="true"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ <dim>48</dim>
+ </port>
+ <port id="1" precision="I64">
+ <dim>2</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="mbox_loc1/out/conv/flat" precision="FP16">
+ <dim>1</dim>
+ <dim>32640</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="717" name="1215" type="Const" version="opset1">
+ <data element_type="f16" offset="1335040" shape="24, 256, 3, 3" size="110592"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>24</dim>
+ <dim>256</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="718" name="mbox_conf1/out/conv/WithoutBiases" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="1, 1" pads_end="1, 1" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>24</dim>
+ <dim>256</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP16">
+ <dim>1</dim>
+ <dim>24</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="719" name="mbox_conf1/out/conv/Dims13831" type="Const" version="opset1">
+ <data element_type="f16" offset="1445632" shape="1, 24, 1, 1" size="48"/>
+ <output>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>24</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="720" name="mbox_conf1/out/conv" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>24</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>1</dim>
+ <dim>24</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="mbox_conf1/out/conv" precision="FP16">
+ <dim>1</dim>
+ <dim>24</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="721" name="1297" type="Const" version="opset1">
+ <data element_type="i64" offset="1334992" shape="4" size="32"/>
+ <output>
+ <port id="0" precision="I64">
+ <dim>4</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="722" name="mbox_conf1/out/conv/perm" type="Transpose" version="opset1">
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>24</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="I64">
+ <dim>4</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="mbox_conf1/out/conv/perm" precision="FP16">
+ <dim>1</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ <dim>24</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="723" name="1303/shapes_concat" type="Const" version="opset1">
+ <data element_type="i64" offset="1335024" shape="2" size="16"/>
+ <output>
+ <port id="0" precision="I64">
+ <dim>2</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="724" name="mbox_conf1/out/conv/flat" type="Reshape" version="opset1">
+ <data special_zero="true"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ <dim>24</dim>
+ </port>
+ <port id="1" precision="I64">
+ <dim>2</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="mbox_conf1/out/conv/flat" precision="FP16">
+ <dim>1</dim>
+ <dim>16320</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="725" name="1295" type="Const" version="opset1">
+ <data element_type="i64" offset="1445680" shape="3" size="24"/>
+ <output>
+ <port id="0" precision="I64">
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="726" name="mbox_conf1/out/conv/flat/reshape" type="Reshape" version="opset1">
+ <data special_zero="true"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>16320</dim>
+ </port>
+ <port id="1" precision="I64">
+ <dim>3</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="mbox_conf1/out/conv/flat/reshape" precision="FP16">
+ <dim>1</dim>
+ <dim>8160</dim>
+ <dim>2</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="727" name="mbox_conf1/out/conv/flat/softmax" type="SoftMax" version="opset1">
+ <data axis="2"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>8160</dim>
+ <dim>2</dim>
+ </port>
+ </input>
+ <output>
+ <port id="1" names="mbox_conf1/out/conv/flat/softmax" precision="FP16">
+ <dim>1</dim>
+ <dim>8160</dim>
+ <dim>2</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="728" name="1298/shapes_concat" type="Const" version="opset1">
+ <data element_type="i64" offset="1335024" shape="2" size="16"/>
+ <output>
+ <port id="0" precision="I64">
+ <dim>2</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="729" name="mbox_conf1/out/conv/flat/softmax/flat" type="Reshape" version="opset1">
+ <data special_zero="true"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>8160</dim>
+ <dim>2</dim>
+ </port>
+ <port id="1" precision="I64">
+ <dim>2</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="mbox_conf1/out/conv/flat/softmax/flat" precision="FP16">
+ <dim>1</dim>
+ <dim>16320</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="730" name="mbox1/priorbox/0_port" type="ShapeOf" version="opset3">
+ <data output_type="i64"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </input>
+ <output>
+ <port id="1" precision="I64">
+ <dim>4</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="731" name="mbox1/priorbox/ss_begin2978639773" type="Const" version="opset1">
+ <data element_type="i64" offset="1445704" shape="1" size="8"/>
+ <output>
+ <port id="0" precision="I64">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="732" name="mbox1/priorbox/ss_end2978740472" type="Const" version="opset1">
+ <data element_type="i64" offset="1445712" shape="1" size="8"/>
+ <output>
+ <port id="0" precision="I64">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="733" name="mbox1/priorbox/ss_stride2978840646" type="Const" version="opset1">
+ <data element_type="i64" offset="1445720" shape="1" size="8"/>
+ <output>
+ <port id="0" precision="I64">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="734" name="mbox1/priorbox/ss_0_port" type="StridedSlice" version="opset1">
+ <data begin_mask="0" ellipsis_mask="0" end_mask="1" new_axis_mask="0" shrink_axis_mask="0"/>
+ <input>
+ <port id="0" precision="I64">
+ <dim>4</dim>
+ </port>
+ <port id="1" precision="I64">
+ <dim>1</dim>
+ </port>
+ <port id="2" precision="I64">
+ <dim>1</dim>
+ </port>
+ <port id="3" precision="I64">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="4" precision="I64">
+ <dim>2</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="735" name="mbox1/priorbox/1_port" type="ShapeOf" version="opset3">
+ <data output_type="i64"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>320</dim>
+ <dim>544</dim>
+ </port>
+ </input>
+ <output>
+ <port id="1" precision="I64">
+ <dim>4</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="736" name="mbox1/priorbox/ss_begin2978640022" type="Const" version="opset1">
+ <data element_type="i64" offset="1445704" shape="1" size="8"/>
+ <output>
+ <port id="0" precision="I64">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="737" name="mbox1/priorbox/ss_end2978740553" type="Const" version="opset1">
+ <data element_type="i64" offset="1445712" shape="1" size="8"/>
+ <output>
+ <port id="0" precision="I64">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="738" name="mbox1/priorbox/ss_stride2978840223" type="Const" version="opset1">
+ <data element_type="i64" offset="1445720" shape="1" size="8"/>
+ <output>
+ <port id="0" precision="I64">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="739" name="mbox1/priorbox/ss_1_port" type="StridedSlice" version="opset1">
+ <data begin_mask="0" ellipsis_mask="0" end_mask="1" new_axis_mask="0" shrink_axis_mask="0"/>
+ <input>
+ <port id="0" precision="I64">
+ <dim>4</dim>
+ </port>
+ <port id="1" precision="I64">
+ <dim>1</dim>
+ </port>
+ <port id="2" precision="I64">
+ <dim>1</dim>
+ </port>
+ <port id="3" precision="I64">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="4" precision="I64">
+ <dim>2</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="740" name="mbox1/priorbox/naked_not_unsqueezed" type="PriorBoxClustered" version="opset1">
+ <data clip="false" height="34.07, 47.11, 54.22, 65.78, 75.56, 80.89, 89.78, 99.26, 115.56, 163.26, 194.07, 197.33" offset="0.5" step="0" step_h="16" step_w="16" variance="0.1, 0.1, 0.2, 0.2" width="11.33, 17, 20.68, 23.52, 28.05, 37.4, 30.03, 35.7, 44.2, 55.25, 78.12, 135.15"/>
+ <input>
+ <port id="0" precision="I64">
+ <dim>2</dim>
+ </port>
+ <port id="1" precision="I64">
+ <dim>2</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP32">
+ <dim>2</dim>
+ <dim>32640</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="741" name="mbox1/priorbox/unsqueeze/value2979640301" type="Const" version="opset1">
+ <data element_type="i64" offset="1445728" shape="1" size="8"/>
+ <output>
+ <port id="0" precision="I64">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="742" name="mbox1/priorbox" type="Unsqueeze" version="opset1">
+ <input>
+ <port id="0" precision="FP32">
+ <dim>2</dim>
+ <dim>32640</dim>
+ </port>
+ <port id="1" precision="I64">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="mbox1/priorbox" precision="FP32">
+ <dim>1</dim>
+ <dim>2</dim>
+ <dim>32640</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="743" name="detection_out" type="DetectionOutput" version="opset1">
+ <data background_label_id="0" clip_after_nms="false" clip_before_nms="false" code_type="caffe.PriorBoxParameter.CENTER_SIZE" confidence_threshold="0.0099999997764825821" decrease_label_id="false" input_height="1" input_width="1" keep_top_k="200" nms_threshold="0.44999998807907104" normalized="true" num_classes="2" objectness_score="0" share_location="true" top_k="400" variance_encoded_in_target="false"/>
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>32640</dim>
+ </port>
+ <port id="1" precision="FP16">
+ <dim>1</dim>
+ <dim>16320</dim>
+ </port>
+ <port id="2" precision="FP32">
+ <dim>1</dim>
+ <dim>2</dim>
+ <dim>32640</dim>
+ </port>
+ </input>
+ <output>
+ <port id="3" names="detection_out" precision="FP16">
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>200</dim>
+ <dim>7</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="744" name="detection_out/sink_port_0" type="Result" version="opset1">
+ <input>
+ <port id="0" precision="FP16">
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>200</dim>
+ <dim>7</dim>
+ </port>
+ </input>
+ </layer>
+ </layers>
+ <edges>
+ <edge from-layer="0" from-port="0" to-layer="735" to-port="0"/>
+ <edge from-layer="0" from-port="0" to-layer="2" to-port="0"/>
+ <edge from-layer="1" from-port="0" to-layer="2" to-port="1"/>
+ <edge from-layer="2" from-port="2" to-layer="4" to-port="0"/>
+ <edge from-layer="3" from-port="0" to-layer="4" to-port="1"/>
+ <edge from-layer="4" from-port="2" to-layer="6" to-port="0"/>
+ <edge from-layer="5" from-port="0" to-layer="6" to-port="1"/>
+ <edge from-layer="6" from-port="2" to-layer="8" to-port="0"/>
+ <edge from-layer="7" from-port="0" to-layer="8" to-port="1"/>
+ <edge from-layer="8" from-port="2" to-layer="9" to-port="0"/>
+ <edge from-layer="9" from-port="1" to-layer="11" to-port="0"/>
+ <edge from-layer="9" from-port="1" to-layer="26" to-port="0"/>
+ <edge from-layer="10" from-port="0" to-layer="11" to-port="1"/>
+ <edge from-layer="11" from-port="2" to-layer="13" to-port="0"/>
+ <edge from-layer="12" from-port="0" to-layer="13" to-port="1"/>
+ <edge from-layer="13" from-port="2" to-layer="15" to-port="0"/>
+ <edge from-layer="14" from-port="0" to-layer="15" to-port="1"/>
+ <edge from-layer="15" from-port="2" to-layer="17" to-port="0"/>
+ <edge from-layer="16" from-port="0" to-layer="17" to-port="1"/>
+ <edge from-layer="17" from-port="2" to-layer="19" to-port="0"/>
+ <edge from-layer="18" from-port="0" to-layer="19" to-port="1"/>
+ <edge from-layer="19" from-port="2" to-layer="21" to-port="0"/>
+ <edge from-layer="20" from-port="0" to-layer="21" to-port="1"/>
+ <edge from-layer="21" from-port="2" to-layer="23" to-port="0"/>
+ <edge from-layer="22" from-port="0" to-layer="23" to-port="1"/>
+ <edge from-layer="23" from-port="2" to-layer="25" to-port="0"/>
+ <edge from-layer="24" from-port="0" to-layer="25" to-port="1"/>
+ <edge from-layer="25" from-port="2" to-layer="26" to-port="1"/>
+ <edge from-layer="26" from-port="2" to-layer="28" to-port="0"/>
+ <edge from-layer="27" from-port="0" to-layer="28" to-port="1"/>
+ <edge from-layer="28" from-port="2" to-layer="30" to-port="0"/>
+ <edge from-layer="28" from-port="2" to-layer="45" to-port="0"/>
+ <edge from-layer="29" from-port="0" to-layer="30" to-port="1"/>
+ <edge from-layer="30" from-port="2" to-layer="32" to-port="0"/>
+ <edge from-layer="31" from-port="0" to-layer="32" to-port="1"/>
+ <edge from-layer="32" from-port="2" to-layer="34" to-port="0"/>
+ <edge from-layer="33" from-port="0" to-layer="34" to-port="1"/>
+ <edge from-layer="34" from-port="2" to-layer="36" to-port="0"/>
+ <edge from-layer="35" from-port="0" to-layer="36" to-port="1"/>
+ <edge from-layer="36" from-port="2" to-layer="38" to-port="0"/>
+ <edge from-layer="37" from-port="0" to-layer="38" to-port="1"/>
+ <edge from-layer="38" from-port="2" to-layer="40" to-port="0"/>
+ <edge from-layer="39" from-port="0" to-layer="40" to-port="1"/>
+ <edge from-layer="40" from-port="2" to-layer="42" to-port="0"/>
+ <edge from-layer="41" from-port="0" to-layer="42" to-port="1"/>
+ <edge from-layer="42" from-port="2" to-layer="44" to-port="0"/>
+ <edge from-layer="43" from-port="0" to-layer="44" to-port="1"/>
+ <edge from-layer="44" from-port="2" to-layer="45" to-port="1"/>
+ <edge from-layer="45" from-port="2" to-layer="47" to-port="0"/>
+ <edge from-layer="46" from-port="0" to-layer="47" to-port="1"/>
+ <edge from-layer="47" from-port="2" to-layer="49" to-port="0"/>
+ <edge from-layer="47" from-port="2" to-layer="64" to-port="0"/>
+ <edge from-layer="48" from-port="0" to-layer="49" to-port="1"/>
+ <edge from-layer="49" from-port="2" to-layer="51" to-port="0"/>
+ <edge from-layer="50" from-port="0" to-layer="51" to-port="1"/>
+ <edge from-layer="51" from-port="2" to-layer="53" to-port="0"/>
+ <edge from-layer="52" from-port="0" to-layer="53" to-port="1"/>
+ <edge from-layer="53" from-port="2" to-layer="55" to-port="0"/>
+ <edge from-layer="54" from-port="0" to-layer="55" to-port="1"/>
+ <edge from-layer="55" from-port="2" to-layer="57" to-port="0"/>
+ <edge from-layer="56" from-port="0" to-layer="57" to-port="1"/>
+ <edge from-layer="57" from-port="2" to-layer="59" to-port="0"/>
+ <edge from-layer="58" from-port="0" to-layer="59" to-port="1"/>
+ <edge from-layer="59" from-port="2" to-layer="61" to-port="0"/>
+ <edge from-layer="60" from-port="0" to-layer="61" to-port="1"/>
+ <edge from-layer="61" from-port="2" to-layer="63" to-port="0"/>
+ <edge from-layer="62" from-port="0" to-layer="63" to-port="1"/>
+ <edge from-layer="63" from-port="2" to-layer="64" to-port="1"/>
+ <edge from-layer="64" from-port="2" to-layer="66" to-port="0"/>
+ <edge from-layer="65" from-port="0" to-layer="66" to-port="1"/>
+ <edge from-layer="66" from-port="2" to-layer="68" to-port="0"/>
+ <edge from-layer="66" from-port="2" to-layer="83" to-port="0"/>
+ <edge from-layer="67" from-port="0" to-layer="68" to-port="1"/>
+ <edge from-layer="68" from-port="2" to-layer="70" to-port="0"/>
+ <edge from-layer="69" from-port="0" to-layer="70" to-port="1"/>
+ <edge from-layer="70" from-port="2" to-layer="72" to-port="0"/>
+ <edge from-layer="71" from-port="0" to-layer="72" to-port="1"/>
+ <edge from-layer="72" from-port="2" to-layer="74" to-port="0"/>
+ <edge from-layer="73" from-port="0" to-layer="74" to-port="1"/>
+ <edge from-layer="74" from-port="2" to-layer="76" to-port="0"/>
+ <edge from-layer="75" from-port="0" to-layer="76" to-port="1"/>
+ <edge from-layer="76" from-port="2" to-layer="78" to-port="0"/>
+ <edge from-layer="77" from-port="0" to-layer="78" to-port="1"/>
+ <edge from-layer="78" from-port="2" to-layer="80" to-port="0"/>
+ <edge from-layer="79" from-port="0" to-layer="80" to-port="1"/>
+ <edge from-layer="80" from-port="2" to-layer="82" to-port="0"/>
+ <edge from-layer="81" from-port="0" to-layer="82" to-port="1"/>
+ <edge from-layer="82" from-port="2" to-layer="83" to-port="1"/>
+ <edge from-layer="83" from-port="2" to-layer="85" to-port="0"/>
+ <edge from-layer="84" from-port="0" to-layer="85" to-port="1"/>
+ <edge from-layer="85" from-port="2" to-layer="86" to-port="0"/>
+ <edge from-layer="85" from-port="2" to-layer="92" to-port="0"/>
+ <edge from-layer="86" from-port="1" to-layer="88" to-port="0"/>
+ <edge from-layer="87" from-port="0" to-layer="88" to-port="1"/>
+ <edge from-layer="88" from-port="2" to-layer="90" to-port="0"/>
+ <edge from-layer="89" from-port="0" to-layer="90" to-port="1"/>
+ <edge from-layer="90" from-port="2" to-layer="107" to-port="0"/>
+ <edge from-layer="91" from-port="0" to-layer="92" to-port="1"/>
+ <edge from-layer="92" from-port="2" to-layer="94" to-port="0"/>
+ <edge from-layer="93" from-port="0" to-layer="94" to-port="1"/>
+ <edge from-layer="94" from-port="2" to-layer="96" to-port="0"/>
+ <edge from-layer="95" from-port="0" to-layer="96" to-port="1"/>
+ <edge from-layer="96" from-port="2" to-layer="98" to-port="0"/>
+ <edge from-layer="97" from-port="0" to-layer="98" to-port="1"/>
+ <edge from-layer="98" from-port="2" to-layer="100" to-port="0"/>
+ <edge from-layer="99" from-port="0" to-layer="100" to-port="1"/>
+ <edge from-layer="100" from-port="2" to-layer="102" to-port="0"/>
+ <edge from-layer="101" from-port="0" to-layer="102" to-port="1"/>
+ <edge from-layer="102" from-port="2" to-layer="104" to-port="0"/>
+ <edge from-layer="103" from-port="0" to-layer="104" to-port="1"/>
+ <edge from-layer="104" from-port="2" to-layer="106" to-port="0"/>
+ <edge from-layer="105" from-port="0" to-layer="106" to-port="1"/>
+ <edge from-layer="106" from-port="2" to-layer="107" to-port="1"/>
+ <edge from-layer="107" from-port="2" to-layer="109" to-port="0"/>
+ <edge from-layer="108" from-port="0" to-layer="109" to-port="1"/>
+ <edge from-layer="109" from-port="2" to-layer="126" to-port="0"/>
+ <edge from-layer="109" from-port="2" to-layer="111" to-port="0"/>
+ <edge from-layer="110" from-port="0" to-layer="111" to-port="1"/>
+ <edge from-layer="111" from-port="2" to-layer="113" to-port="0"/>
+ <edge from-layer="112" from-port="0" to-layer="113" to-port="1"/>
+ <edge from-layer="113" from-port="2" to-layer="115" to-port="0"/>
+ <edge from-layer="114" from-port="0" to-layer="115" to-port="1"/>
+ <edge from-layer="115" from-port="2" to-layer="117" to-port="0"/>
+ <edge from-layer="116" from-port="0" to-layer="117" to-port="1"/>
+ <edge from-layer="117" from-port="2" to-layer="119" to-port="0"/>
+ <edge from-layer="118" from-port="0" to-layer="119" to-port="1"/>
+ <edge from-layer="119" from-port="2" to-layer="121" to-port="0"/>
+ <edge from-layer="120" from-port="0" to-layer="121" to-port="1"/>
+ <edge from-layer="121" from-port="2" to-layer="123" to-port="0"/>
+ <edge from-layer="122" from-port="0" to-layer="123" to-port="1"/>
+ <edge from-layer="123" from-port="2" to-layer="125" to-port="0"/>
+ <edge from-layer="124" from-port="0" to-layer="125" to-port="1"/>
+ <edge from-layer="125" from-port="2" to-layer="126" to-port="1"/>
+ <edge from-layer="126" from-port="2" to-layer="128" to-port="0"/>
+ <edge from-layer="127" from-port="0" to-layer="128" to-port="1"/>
+ <edge from-layer="128" from-port="2" to-layer="130" to-port="0"/>
+ <edge from-layer="128" from-port="2" to-layer="145" to-port="0"/>
+ <edge from-layer="129" from-port="0" to-layer="130" to-port="1"/>
+ <edge from-layer="130" from-port="2" to-layer="132" to-port="0"/>
+ <edge from-layer="131" from-port="0" to-layer="132" to-port="1"/>
+ <edge from-layer="132" from-port="2" to-layer="134" to-port="0"/>
+ <edge from-layer="133" from-port="0" to-layer="134" to-port="1"/>
+ <edge from-layer="134" from-port="2" to-layer="136" to-port="0"/>
+ <edge from-layer="135" from-port="0" to-layer="136" to-port="1"/>
+ <edge from-layer="136" from-port="2" to-layer="138" to-port="0"/>
+ <edge from-layer="137" from-port="0" to-layer="138" to-port="1"/>
+ <edge from-layer="138" from-port="2" to-layer="140" to-port="0"/>
+ <edge from-layer="139" from-port="0" to-layer="140" to-port="1"/>
+ <edge from-layer="140" from-port="2" to-layer="142" to-port="0"/>
+ <edge from-layer="141" from-port="0" to-layer="142" to-port="1"/>
+ <edge from-layer="142" from-port="2" to-layer="144" to-port="0"/>
+ <edge from-layer="143" from-port="0" to-layer="144" to-port="1"/>
+ <edge from-layer="144" from-port="2" to-layer="145" to-port="1"/>
+ <edge from-layer="145" from-port="2" to-layer="147" to-port="0"/>
+ <edge from-layer="146" from-port="0" to-layer="147" to-port="1"/>
+ <edge from-layer="147" from-port="2" to-layer="149" to-port="0"/>
+ <edge from-layer="147" from-port="2" to-layer="164" to-port="0"/>
+ <edge from-layer="148" from-port="0" to-layer="149" to-port="1"/>
+ <edge from-layer="149" from-port="2" to-layer="151" to-port="0"/>
+ <edge from-layer="150" from-port="0" to-layer="151" to-port="1"/>
+ <edge from-layer="151" from-port="2" to-layer="153" to-port="0"/>
+ <edge from-layer="152" from-port="0" to-layer="153" to-port="1"/>
+ <edge from-layer="153" from-port="2" to-layer="155" to-port="0"/>
+ <edge from-layer="154" from-port="0" to-layer="155" to-port="1"/>
+ <edge from-layer="155" from-port="2" to-layer="157" to-port="0"/>
+ <edge from-layer="156" from-port="0" to-layer="157" to-port="1"/>
+ <edge from-layer="157" from-port="2" to-layer="159" to-port="0"/>
+ <edge from-layer="158" from-port="0" to-layer="159" to-port="1"/>
+ <edge from-layer="159" from-port="2" to-layer="161" to-port="0"/>
+ <edge from-layer="160" from-port="0" to-layer="161" to-port="1"/>
+ <edge from-layer="161" from-port="2" to-layer="163" to-port="0"/>
+ <edge from-layer="162" from-port="0" to-layer="163" to-port="1"/>
+ <edge from-layer="163" from-port="2" to-layer="164" to-port="1"/>
+ <edge from-layer="164" from-port="2" to-layer="166" to-port="0"/>
+ <edge from-layer="165" from-port="0" to-layer="166" to-port="1"/>
+ <edge from-layer="166" from-port="2" to-layer="168" to-port="0"/>
+ <edge from-layer="166" from-port="2" to-layer="183" to-port="0"/>
+ <edge from-layer="167" from-port="0" to-layer="168" to-port="1"/>
+ <edge from-layer="168" from-port="2" to-layer="170" to-port="0"/>
+ <edge from-layer="169" from-port="0" to-layer="170" to-port="1"/>
+ <edge from-layer="170" from-port="2" to-layer="172" to-port="0"/>
+ <edge from-layer="171" from-port="0" to-layer="172" to-port="1"/>
+ <edge from-layer="172" from-port="2" to-layer="174" to-port="0"/>
+ <edge from-layer="173" from-port="0" to-layer="174" to-port="1"/>
+ <edge from-layer="174" from-port="2" to-layer="176" to-port="0"/>
+ <edge from-layer="175" from-port="0" to-layer="176" to-port="1"/>
+ <edge from-layer="176" from-port="2" to-layer="178" to-port="0"/>
+ <edge from-layer="177" from-port="0" to-layer="178" to-port="1"/>
+ <edge from-layer="178" from-port="2" to-layer="180" to-port="0"/>
+ <edge from-layer="179" from-port="0" to-layer="180" to-port="1"/>
+ <edge from-layer="180" from-port="2" to-layer="182" to-port="0"/>
+ <edge from-layer="181" from-port="0" to-layer="182" to-port="1"/>
+ <edge from-layer="182" from-port="2" to-layer="183" to-port="1"/>
+ <edge from-layer="183" from-port="2" to-layer="185" to-port="0"/>
+ <edge from-layer="184" from-port="0" to-layer="185" to-port="1"/>
+ <edge from-layer="185" from-port="2" to-layer="187" to-port="0"/>
+ <edge from-layer="185" from-port="2" to-layer="202" to-port="0"/>
+ <edge from-layer="186" from-port="0" to-layer="187" to-port="1"/>
+ <edge from-layer="187" from-port="2" to-layer="189" to-port="0"/>
+ <edge from-layer="188" from-port="0" to-layer="189" to-port="1"/>
+ <edge from-layer="189" from-port="2" to-layer="191" to-port="0"/>
+ <edge from-layer="190" from-port="0" to-layer="191" to-port="1"/>
+ <edge from-layer="191" from-port="2" to-layer="193" to-port="0"/>
+ <edge from-layer="192" from-port="0" to-layer="193" to-port="1"/>
+ <edge from-layer="193" from-port="2" to-layer="195" to-port="0"/>
+ <edge from-layer="194" from-port="0" to-layer="195" to-port="1"/>
+ <edge from-layer="195" from-port="2" to-layer="197" to-port="0"/>
+ <edge from-layer="196" from-port="0" to-layer="197" to-port="1"/>
+ <edge from-layer="197" from-port="2" to-layer="199" to-port="0"/>
+ <edge from-layer="198" from-port="0" to-layer="199" to-port="1"/>
+ <edge from-layer="199" from-port="2" to-layer="201" to-port="0"/>
+ <edge from-layer="200" from-port="0" to-layer="201" to-port="1"/>
+ <edge from-layer="201" from-port="2" to-layer="202" to-port="1"/>
+ <edge from-layer="202" from-port="2" to-layer="204" to-port="0"/>
+ <edge from-layer="203" from-port="0" to-layer="204" to-port="1"/>
+ <edge from-layer="204" from-port="2" to-layer="221" to-port="0"/>
+ <edge from-layer="204" from-port="2" to-layer="206" to-port="0"/>
+ <edge from-layer="205" from-port="0" to-layer="206" to-port="1"/>
+ <edge from-layer="206" from-port="2" to-layer="208" to-port="0"/>
+ <edge from-layer="207" from-port="0" to-layer="208" to-port="1"/>
+ <edge from-layer="208" from-port="2" to-layer="210" to-port="0"/>
+ <edge from-layer="209" from-port="0" to-layer="210" to-port="1"/>
+ <edge from-layer="210" from-port="2" to-layer="212" to-port="0"/>
+ <edge from-layer="211" from-port="0" to-layer="212" to-port="1"/>
+ <edge from-layer="212" from-port="2" to-layer="214" to-port="0"/>
+ <edge from-layer="213" from-port="0" to-layer="214" to-port="1"/>
+ <edge from-layer="214" from-port="2" to-layer="216" to-port="0"/>
+ <edge from-layer="215" from-port="0" to-layer="216" to-port="1"/>
+ <edge from-layer="216" from-port="2" to-layer="218" to-port="0"/>
+ <edge from-layer="217" from-port="0" to-layer="218" to-port="1"/>
+ <edge from-layer="218" from-port="2" to-layer="220" to-port="0"/>
+ <edge from-layer="219" from-port="0" to-layer="220" to-port="1"/>
+ <edge from-layer="220" from-port="2" to-layer="221" to-port="1"/>
+ <edge from-layer="221" from-port="2" to-layer="223" to-port="0"/>
+ <edge from-layer="222" from-port="0" to-layer="223" to-port="1"/>
+ <edge from-layer="223" from-port="2" to-layer="225" to-port="0"/>
+ <edge from-layer="223" from-port="2" to-layer="240" to-port="0"/>
+ <edge from-layer="224" from-port="0" to-layer="225" to-port="1"/>
+ <edge from-layer="225" from-port="2" to-layer="227" to-port="0"/>
+ <edge from-layer="226" from-port="0" to-layer="227" to-port="1"/>
+ <edge from-layer="227" from-port="2" to-layer="229" to-port="0"/>
+ <edge from-layer="228" from-port="0" to-layer="229" to-port="1"/>
+ <edge from-layer="229" from-port="2" to-layer="231" to-port="0"/>
+ <edge from-layer="230" from-port="0" to-layer="231" to-port="1"/>
+ <edge from-layer="231" from-port="2" to-layer="233" to-port="0"/>
+ <edge from-layer="232" from-port="0" to-layer="233" to-port="1"/>
+ <edge from-layer="233" from-port="2" to-layer="235" to-port="0"/>
+ <edge from-layer="234" from-port="0" to-layer="235" to-port="1"/>
+ <edge from-layer="235" from-port="2" to-layer="237" to-port="0"/>
+ <edge from-layer="236" from-port="0" to-layer="237" to-port="1"/>
+ <edge from-layer="237" from-port="2" to-layer="239" to-port="0"/>
+ <edge from-layer="238" from-port="0" to-layer="239" to-port="1"/>
+ <edge from-layer="239" from-port="2" to-layer="240" to-port="1"/>
+ <edge from-layer="240" from-port="2" to-layer="242" to-port="0"/>
+ <edge from-layer="241" from-port="0" to-layer="242" to-port="1"/>
+ <edge from-layer="242" from-port="2" to-layer="244" to-port="0"/>
+ <edge from-layer="242" from-port="2" to-layer="259" to-port="0"/>
+ <edge from-layer="243" from-port="0" to-layer="244" to-port="1"/>
+ <edge from-layer="244" from-port="2" to-layer="246" to-port="0"/>
+ <edge from-layer="245" from-port="0" to-layer="246" to-port="1"/>
+ <edge from-layer="246" from-port="2" to-layer="248" to-port="0"/>
+ <edge from-layer="247" from-port="0" to-layer="248" to-port="1"/>
+ <edge from-layer="248" from-port="2" to-layer="250" to-port="0"/>
+ <edge from-layer="249" from-port="0" to-layer="250" to-port="1"/>
+ <edge from-layer="250" from-port="2" to-layer="252" to-port="0"/>
+ <edge from-layer="251" from-port="0" to-layer="252" to-port="1"/>
+ <edge from-layer="252" from-port="2" to-layer="254" to-port="0"/>
+ <edge from-layer="253" from-port="0" to-layer="254" to-port="1"/>
+ <edge from-layer="254" from-port="2" to-layer="256" to-port="0"/>
+ <edge from-layer="255" from-port="0" to-layer="256" to-port="1"/>
+ <edge from-layer="256" from-port="2" to-layer="258" to-port="0"/>
+ <edge from-layer="257" from-port="0" to-layer="258" to-port="1"/>
+ <edge from-layer="258" from-port="2" to-layer="259" to-port="1"/>
+ <edge from-layer="259" from-port="2" to-layer="261" to-port="0"/>
+ <edge from-layer="260" from-port="0" to-layer="261" to-port="1"/>
+ <edge from-layer="261" from-port="2" to-layer="262" to-port="0"/>
+ <edge from-layer="261" from-port="2" to-layer="268" to-port="0"/>
+ <edge from-layer="262" from-port="1" to-layer="264" to-port="0"/>
+ <edge from-layer="263" from-port="0" to-layer="264" to-port="1"/>
+ <edge from-layer="264" from-port="2" to-layer="266" to-port="0"/>
+ <edge from-layer="265" from-port="0" to-layer="266" to-port="1"/>
+ <edge from-layer="266" from-port="2" to-layer="283" to-port="0"/>
+ <edge from-layer="267" from-port="0" to-layer="268" to-port="1"/>
+ <edge from-layer="268" from-port="2" to-layer="270" to-port="0"/>
+ <edge from-layer="269" from-port="0" to-layer="270" to-port="1"/>
+ <edge from-layer="270" from-port="2" to-layer="272" to-port="0"/>
+ <edge from-layer="271" from-port="0" to-layer="272" to-port="1"/>
+ <edge from-layer="272" from-port="2" to-layer="274" to-port="0"/>
+ <edge from-layer="273" from-port="0" to-layer="274" to-port="1"/>
+ <edge from-layer="274" from-port="2" to-layer="276" to-port="0"/>
+ <edge from-layer="275" from-port="0" to-layer="276" to-port="1"/>
+ <edge from-layer="276" from-port="2" to-layer="278" to-port="0"/>
+ <edge from-layer="277" from-port="0" to-layer="278" to-port="1"/>
+ <edge from-layer="278" from-port="2" to-layer="280" to-port="0"/>
+ <edge from-layer="279" from-port="0" to-layer="280" to-port="1"/>
+ <edge from-layer="280" from-port="2" to-layer="282" to-port="0"/>
+ <edge from-layer="281" from-port="0" to-layer="282" to-port="1"/>
+ <edge from-layer="282" from-port="2" to-layer="283" to-port="1"/>
+ <edge from-layer="283" from-port="2" to-layer="285" to-port="0"/>
+ <edge from-layer="284" from-port="0" to-layer="285" to-port="1"/>
+ <edge from-layer="285" from-port="2" to-layer="302" to-port="0"/>
+ <edge from-layer="285" from-port="2" to-layer="287" to-port="0"/>
+ <edge from-layer="286" from-port="0" to-layer="287" to-port="1"/>
+ <edge from-layer="287" from-port="2" to-layer="289" to-port="0"/>
+ <edge from-layer="288" from-port="0" to-layer="289" to-port="1"/>
+ <edge from-layer="289" from-port="2" to-layer="291" to-port="0"/>
+ <edge from-layer="290" from-port="0" to-layer="291" to-port="1"/>
+ <edge from-layer="291" from-port="2" to-layer="293" to-port="0"/>
+ <edge from-layer="292" from-port="0" to-layer="293" to-port="1"/>
+ <edge from-layer="293" from-port="2" to-layer="295" to-port="0"/>
+ <edge from-layer="294" from-port="0" to-layer="295" to-port="1"/>
+ <edge from-layer="295" from-port="2" to-layer="297" to-port="0"/>
+ <edge from-layer="296" from-port="0" to-layer="297" to-port="1"/>
+ <edge from-layer="297" from-port="2" to-layer="299" to-port="0"/>
+ <edge from-layer="298" from-port="0" to-layer="299" to-port="1"/>
+ <edge from-layer="299" from-port="2" to-layer="301" to-port="0"/>
+ <edge from-layer="300" from-port="0" to-layer="301" to-port="1"/>
+ <edge from-layer="301" from-port="2" to-layer="302" to-port="1"/>
+ <edge from-layer="302" from-port="2" to-layer="304" to-port="0"/>
+ <edge from-layer="303" from-port="0" to-layer="304" to-port="1"/>
+ <edge from-layer="304" from-port="2" to-layer="306" to-port="0"/>
+ <edge from-layer="304" from-port="2" to-layer="321" to-port="0"/>
+ <edge from-layer="305" from-port="0" to-layer="306" to-port="1"/>
+ <edge from-layer="306" from-port="2" to-layer="308" to-port="0"/>
+ <edge from-layer="307" from-port="0" to-layer="308" to-port="1"/>
+ <edge from-layer="308" from-port="2" to-layer="310" to-port="0"/>
+ <edge from-layer="309" from-port="0" to-layer="310" to-port="1"/>
+ <edge from-layer="310" from-port="2" to-layer="312" to-port="0"/>
+ <edge from-layer="311" from-port="0" to-layer="312" to-port="1"/>
+ <edge from-layer="312" from-port="2" to-layer="314" to-port="0"/>
+ <edge from-layer="313" from-port="0" to-layer="314" to-port="1"/>
+ <edge from-layer="314" from-port="2" to-layer="316" to-port="0"/>
+ <edge from-layer="315" from-port="0" to-layer="316" to-port="1"/>
+ <edge from-layer="316" from-port="2" to-layer="318" to-port="0"/>
+ <edge from-layer="317" from-port="0" to-layer="318" to-port="1"/>
+ <edge from-layer="318" from-port="2" to-layer="320" to-port="0"/>
+ <edge from-layer="319" from-port="0" to-layer="320" to-port="1"/>
+ <edge from-layer="320" from-port="2" to-layer="321" to-port="1"/>
+ <edge from-layer="321" from-port="2" to-layer="323" to-port="0"/>
+ <edge from-layer="322" from-port="0" to-layer="323" to-port="1"/>
+ <edge from-layer="323" from-port="2" to-layer="325" to-port="0"/>
+ <edge from-layer="323" from-port="2" to-layer="340" to-port="0"/>
+ <edge from-layer="324" from-port="0" to-layer="325" to-port="1"/>
+ <edge from-layer="325" from-port="2" to-layer="327" to-port="0"/>
+ <edge from-layer="326" from-port="0" to-layer="327" to-port="1"/>
+ <edge from-layer="327" from-port="2" to-layer="329" to-port="0"/>
+ <edge from-layer="328" from-port="0" to-layer="329" to-port="1"/>
+ <edge from-layer="329" from-port="2" to-layer="331" to-port="0"/>
+ <edge from-layer="330" from-port="0" to-layer="331" to-port="1"/>
+ <edge from-layer="331" from-port="2" to-layer="333" to-port="0"/>
+ <edge from-layer="332" from-port="0" to-layer="333" to-port="1"/>
+ <edge from-layer="333" from-port="2" to-layer="335" to-port="0"/>
+ <edge from-layer="334" from-port="0" to-layer="335" to-port="1"/>
+ <edge from-layer="335" from-port="2" to-layer="337" to-port="0"/>
+ <edge from-layer="336" from-port="0" to-layer="337" to-port="1"/>
+ <edge from-layer="337" from-port="2" to-layer="339" to-port="0"/>
+ <edge from-layer="338" from-port="0" to-layer="339" to-port="1"/>
+ <edge from-layer="339" from-port="2" to-layer="340" to-port="1"/>
+ <edge from-layer="340" from-port="2" to-layer="342" to-port="0"/>
+ <edge from-layer="341" from-port="0" to-layer="342" to-port="1"/>
+ <edge from-layer="342" from-port="2" to-layer="359" to-port="0"/>
+ <edge from-layer="342" from-port="2" to-layer="344" to-port="0"/>
+ <edge from-layer="343" from-port="0" to-layer="344" to-port="1"/>
+ <edge from-layer="344" from-port="2" to-layer="346" to-port="0"/>
+ <edge from-layer="345" from-port="0" to-layer="346" to-port="1"/>
+ <edge from-layer="346" from-port="2" to-layer="348" to-port="0"/>
+ <edge from-layer="347" from-port="0" to-layer="348" to-port="1"/>
+ <edge from-layer="348" from-port="2" to-layer="350" to-port="0"/>
+ <edge from-layer="349" from-port="0" to-layer="350" to-port="1"/>
+ <edge from-layer="350" from-port="2" to-layer="352" to-port="0"/>
+ <edge from-layer="351" from-port="0" to-layer="352" to-port="1"/>
+ <edge from-layer="352" from-port="2" to-layer="354" to-port="0"/>
+ <edge from-layer="353" from-port="0" to-layer="354" to-port="1"/>
+ <edge from-layer="354" from-port="2" to-layer="356" to-port="0"/>
+ <edge from-layer="355" from-port="0" to-layer="356" to-port="1"/>
+ <edge from-layer="356" from-port="2" to-layer="358" to-port="0"/>
+ <edge from-layer="357" from-port="0" to-layer="358" to-port="1"/>
+ <edge from-layer="358" from-port="2" to-layer="359" to-port="1"/>
+ <edge from-layer="359" from-port="2" to-layer="361" to-port="0"/>
+ <edge from-layer="360" from-port="0" to-layer="361" to-port="1"/>
+ <edge from-layer="361" from-port="2" to-layer="378" to-port="0"/>
+ <edge from-layer="361" from-port="2" to-layer="363" to-port="0"/>
+ <edge from-layer="362" from-port="0" to-layer="363" to-port="1"/>
+ <edge from-layer="363" from-port="2" to-layer="365" to-port="0"/>
+ <edge from-layer="364" from-port="0" to-layer="365" to-port="1"/>
+ <edge from-layer="365" from-port="2" to-layer="367" to-port="0"/>
+ <edge from-layer="366" from-port="0" to-layer="367" to-port="1"/>
+ <edge from-layer="367" from-port="2" to-layer="369" to-port="0"/>
+ <edge from-layer="368" from-port="0" to-layer="369" to-port="1"/>
+ <edge from-layer="369" from-port="2" to-layer="371" to-port="0"/>
+ <edge from-layer="370" from-port="0" to-layer="371" to-port="1"/>
+ <edge from-layer="371" from-port="2" to-layer="373" to-port="0"/>
+ <edge from-layer="372" from-port="0" to-layer="373" to-port="1"/>
+ <edge from-layer="373" from-port="2" to-layer="375" to-port="0"/>
+ <edge from-layer="374" from-port="0" to-layer="375" to-port="1"/>
+ <edge from-layer="375" from-port="2" to-layer="377" to-port="0"/>
+ <edge from-layer="376" from-port="0" to-layer="377" to-port="1"/>
+ <edge from-layer="377" from-port="2" to-layer="378" to-port="1"/>
+ <edge from-layer="378" from-port="2" to-layer="380" to-port="0"/>
+ <edge from-layer="379" from-port="0" to-layer="380" to-port="1"/>
+ <edge from-layer="380" from-port="2" to-layer="397" to-port="0"/>
+ <edge from-layer="380" from-port="2" to-layer="382" to-port="0"/>
+ <edge from-layer="381" from-port="0" to-layer="382" to-port="1"/>
+ <edge from-layer="382" from-port="2" to-layer="384" to-port="0"/>
+ <edge from-layer="383" from-port="0" to-layer="384" to-port="1"/>
+ <edge from-layer="384" from-port="2" to-layer="386" to-port="0"/>
+ <edge from-layer="385" from-port="0" to-layer="386" to-port="1"/>
+ <edge from-layer="386" from-port="2" to-layer="388" to-port="0"/>
+ <edge from-layer="387" from-port="0" to-layer="388" to-port="1"/>
+ <edge from-layer="388" from-port="2" to-layer="390" to-port="0"/>
+ <edge from-layer="389" from-port="0" to-layer="390" to-port="1"/>
+ <edge from-layer="390" from-port="2" to-layer="392" to-port="0"/>
+ <edge from-layer="391" from-port="0" to-layer="392" to-port="1"/>
+ <edge from-layer="392" from-port="2" to-layer="394" to-port="0"/>
+ <edge from-layer="393" from-port="0" to-layer="394" to-port="1"/>
+ <edge from-layer="394" from-port="2" to-layer="396" to-port="0"/>
+ <edge from-layer="395" from-port="0" to-layer="396" to-port="1"/>
+ <edge from-layer="396" from-port="2" to-layer="397" to-port="1"/>
+ <edge from-layer="397" from-port="2" to-layer="399" to-port="0"/>
+ <edge from-layer="398" from-port="0" to-layer="399" to-port="1"/>
+ <edge from-layer="399" from-port="2" to-layer="401" to-port="0"/>
+ <edge from-layer="399" from-port="2" to-layer="416" to-port="0"/>
+ <edge from-layer="400" from-port="0" to-layer="401" to-port="1"/>
+ <edge from-layer="401" from-port="2" to-layer="403" to-port="0"/>
+ <edge from-layer="402" from-port="0" to-layer="403" to-port="1"/>
+ <edge from-layer="403" from-port="2" to-layer="405" to-port="0"/>
+ <edge from-layer="404" from-port="0" to-layer="405" to-port="1"/>
+ <edge from-layer="405" from-port="2" to-layer="407" to-port="0"/>
+ <edge from-layer="406" from-port="0" to-layer="407" to-port="1"/>
+ <edge from-layer="407" from-port="2" to-layer="409" to-port="0"/>
+ <edge from-layer="408" from-port="0" to-layer="409" to-port="1"/>
+ <edge from-layer="409" from-port="2" to-layer="411" to-port="0"/>
+ <edge from-layer="410" from-port="0" to-layer="411" to-port="1"/>
+ <edge from-layer="411" from-port="2" to-layer="413" to-port="0"/>
+ <edge from-layer="412" from-port="0" to-layer="413" to-port="1"/>
+ <edge from-layer="413" from-port="2" to-layer="415" to-port="0"/>
+ <edge from-layer="414" from-port="0" to-layer="415" to-port="1"/>
+ <edge from-layer="415" from-port="2" to-layer="416" to-port="1"/>
+ <edge from-layer="416" from-port="2" to-layer="418" to-port="0"/>
+ <edge from-layer="417" from-port="0" to-layer="418" to-port="1"/>
+ <edge from-layer="418" from-port="2" to-layer="420" to-port="0"/>
+ <edge from-layer="418" from-port="2" to-layer="435" to-port="0"/>
+ <edge from-layer="419" from-port="0" to-layer="420" to-port="1"/>
+ <edge from-layer="420" from-port="2" to-layer="422" to-port="0"/>
+ <edge from-layer="421" from-port="0" to-layer="422" to-port="1"/>
+ <edge from-layer="422" from-port="2" to-layer="424" to-port="0"/>
+ <edge from-layer="423" from-port="0" to-layer="424" to-port="1"/>
+ <edge from-layer="424" from-port="2" to-layer="426" to-port="0"/>
+ <edge from-layer="425" from-port="0" to-layer="426" to-port="1"/>
+ <edge from-layer="426" from-port="2" to-layer="428" to-port="0"/>
+ <edge from-layer="427" from-port="0" to-layer="428" to-port="1"/>
+ <edge from-layer="428" from-port="2" to-layer="430" to-port="0"/>
+ <edge from-layer="429" from-port="0" to-layer="430" to-port="1"/>
+ <edge from-layer="430" from-port="2" to-layer="432" to-port="0"/>
+ <edge from-layer="431" from-port="0" to-layer="432" to-port="1"/>
+ <edge from-layer="432" from-port="2" to-layer="434" to-port="0"/>
+ <edge from-layer="433" from-port="0" to-layer="434" to-port="1"/>
+ <edge from-layer="434" from-port="2" to-layer="435" to-port="1"/>
+ <edge from-layer="435" from-port="2" to-layer="437" to-port="0"/>
+ <edge from-layer="436" from-port="0" to-layer="437" to-port="1"/>
+ <edge from-layer="437" from-port="2" to-layer="454" to-port="0"/>
+ <edge from-layer="437" from-port="2" to-layer="439" to-port="0"/>
+ <edge from-layer="438" from-port="0" to-layer="439" to-port="1"/>
+ <edge from-layer="439" from-port="2" to-layer="441" to-port="0"/>
+ <edge from-layer="440" from-port="0" to-layer="441" to-port="1"/>
+ <edge from-layer="441" from-port="2" to-layer="443" to-port="0"/>
+ <edge from-layer="442" from-port="0" to-layer="443" to-port="1"/>
+ <edge from-layer="443" from-port="2" to-layer="445" to-port="0"/>
+ <edge from-layer="444" from-port="0" to-layer="445" to-port="1"/>
+ <edge from-layer="445" from-port="2" to-layer="447" to-port="0"/>
+ <edge from-layer="446" from-port="0" to-layer="447" to-port="1"/>
+ <edge from-layer="447" from-port="2" to-layer="449" to-port="0"/>
+ <edge from-layer="448" from-port="0" to-layer="449" to-port="1"/>
+ <edge from-layer="449" from-port="2" to-layer="451" to-port="0"/>
+ <edge from-layer="450" from-port="0" to-layer="451" to-port="1"/>
+ <edge from-layer="451" from-port="2" to-layer="453" to-port="0"/>
+ <edge from-layer="452" from-port="0" to-layer="453" to-port="1"/>
+ <edge from-layer="453" from-port="2" to-layer="454" to-port="1"/>
+ <edge from-layer="454" from-port="2" to-layer="456" to-port="0"/>
+ <edge from-layer="455" from-port="0" to-layer="456" to-port="1"/>
+ <edge from-layer="456" from-port="2" to-layer="473" to-port="0"/>
+ <edge from-layer="456" from-port="2" to-layer="458" to-port="0"/>
+ <edge from-layer="457" from-port="0" to-layer="458" to-port="1"/>
+ <edge from-layer="458" from-port="2" to-layer="460" to-port="0"/>
+ <edge from-layer="459" from-port="0" to-layer="460" to-port="1"/>
+ <edge from-layer="460" from-port="2" to-layer="462" to-port="0"/>
+ <edge from-layer="461" from-port="0" to-layer="462" to-port="1"/>
+ <edge from-layer="462" from-port="2" to-layer="464" to-port="0"/>
+ <edge from-layer="463" from-port="0" to-layer="464" to-port="1"/>
+ <edge from-layer="464" from-port="2" to-layer="466" to-port="0"/>
+ <edge from-layer="465" from-port="0" to-layer="466" to-port="1"/>
+ <edge from-layer="466" from-port="2" to-layer="468" to-port="0"/>
+ <edge from-layer="467" from-port="0" to-layer="468" to-port="1"/>
+ <edge from-layer="468" from-port="2" to-layer="470" to-port="0"/>
+ <edge from-layer="469" from-port="0" to-layer="470" to-port="1"/>
+ <edge from-layer="470" from-port="2" to-layer="472" to-port="0"/>
+ <edge from-layer="471" from-port="0" to-layer="472" to-port="1"/>
+ <edge from-layer="472" from-port="2" to-layer="473" to-port="1"/>
+ <edge from-layer="473" from-port="2" to-layer="475" to-port="0"/>
+ <edge from-layer="474" from-port="0" to-layer="475" to-port="1"/>
+ <edge from-layer="475" from-port="2" to-layer="476" to-port="0"/>
+ <edge from-layer="475" from-port="2" to-layer="482" to-port="0"/>
+ <edge from-layer="476" from-port="1" to-layer="478" to-port="0"/>
+ <edge from-layer="477" from-port="0" to-layer="478" to-port="1"/>
+ <edge from-layer="478" from-port="2" to-layer="480" to-port="0"/>
+ <edge from-layer="479" from-port="0" to-layer="480" to-port="1"/>
+ <edge from-layer="480" from-port="2" to-layer="497" to-port="0"/>
+ <edge from-layer="481" from-port="0" to-layer="482" to-port="1"/>
+ <edge from-layer="482" from-port="2" to-layer="484" to-port="0"/>
+ <edge from-layer="483" from-port="0" to-layer="484" to-port="1"/>
+ <edge from-layer="484" from-port="2" to-layer="486" to-port="0"/>
+ <edge from-layer="485" from-port="0" to-layer="486" to-port="1"/>
+ <edge from-layer="486" from-port="2" to-layer="488" to-port="0"/>
+ <edge from-layer="487" from-port="0" to-layer="488" to-port="1"/>
+ <edge from-layer="488" from-port="2" to-layer="490" to-port="0"/>
+ <edge from-layer="489" from-port="0" to-layer="490" to-port="1"/>
+ <edge from-layer="490" from-port="2" to-layer="492" to-port="0"/>
+ <edge from-layer="491" from-port="0" to-layer="492" to-port="1"/>
+ <edge from-layer="492" from-port="2" to-layer="494" to-port="0"/>
+ <edge from-layer="493" from-port="0" to-layer="494" to-port="1"/>
+ <edge from-layer="494" from-port="2" to-layer="496" to-port="0"/>
+ <edge from-layer="495" from-port="0" to-layer="496" to-port="1"/>
+ <edge from-layer="496" from-port="2" to-layer="497" to-port="1"/>
+ <edge from-layer="497" from-port="2" to-layer="499" to-port="0"/>
+ <edge from-layer="498" from-port="0" to-layer="499" to-port="1"/>
+ <edge from-layer="499" from-port="2" to-layer="501" to-port="0"/>
+ <edge from-layer="499" from-port="2" to-layer="516" to-port="0"/>
+ <edge from-layer="500" from-port="0" to-layer="501" to-port="1"/>
+ <edge from-layer="501" from-port="2" to-layer="503" to-port="0"/>
+ <edge from-layer="502" from-port="0" to-layer="503" to-port="1"/>
+ <edge from-layer="503" from-port="2" to-layer="505" to-port="0"/>
+ <edge from-layer="504" from-port="0" to-layer="505" to-port="1"/>
+ <edge from-layer="505" from-port="2" to-layer="507" to-port="0"/>
+ <edge from-layer="506" from-port="0" to-layer="507" to-port="1"/>
+ <edge from-layer="507" from-port="2" to-layer="509" to-port="0"/>
+ <edge from-layer="508" from-port="0" to-layer="509" to-port="1"/>
+ <edge from-layer="509" from-port="2" to-layer="511" to-port="0"/>
+ <edge from-layer="510" from-port="0" to-layer="511" to-port="1"/>
+ <edge from-layer="511" from-port="2" to-layer="513" to-port="0"/>
+ <edge from-layer="512" from-port="0" to-layer="513" to-port="1"/>
+ <edge from-layer="513" from-port="2" to-layer="515" to-port="0"/>
+ <edge from-layer="514" from-port="0" to-layer="515" to-port="1"/>
+ <edge from-layer="515" from-port="2" to-layer="516" to-port="1"/>
+ <edge from-layer="516" from-port="2" to-layer="518" to-port="0"/>
+ <edge from-layer="517" from-port="0" to-layer="518" to-port="1"/>
+ <edge from-layer="518" from-port="2" to-layer="520" to-port="0"/>
+ <edge from-layer="518" from-port="2" to-layer="535" to-port="0"/>
+ <edge from-layer="519" from-port="0" to-layer="520" to-port="1"/>
+ <edge from-layer="520" from-port="2" to-layer="522" to-port="0"/>
+ <edge from-layer="521" from-port="0" to-layer="522" to-port="1"/>
+ <edge from-layer="522" from-port="2" to-layer="524" to-port="0"/>
+ <edge from-layer="523" from-port="0" to-layer="524" to-port="1"/>
+ <edge from-layer="524" from-port="2" to-layer="526" to-port="0"/>
+ <edge from-layer="525" from-port="0" to-layer="526" to-port="1"/>
+ <edge from-layer="526" from-port="2" to-layer="528" to-port="0"/>
+ <edge from-layer="527" from-port="0" to-layer="528" to-port="1"/>
+ <edge from-layer="528" from-port="2" to-layer="530" to-port="0"/>
+ <edge from-layer="529" from-port="0" to-layer="530" to-port="1"/>
+ <edge from-layer="530" from-port="2" to-layer="532" to-port="0"/>
+ <edge from-layer="531" from-port="0" to-layer="532" to-port="1"/>
+ <edge from-layer="532" from-port="2" to-layer="534" to-port="0"/>
+ <edge from-layer="533" from-port="0" to-layer="534" to-port="1"/>
+ <edge from-layer="534" from-port="2" to-layer="535" to-port="1"/>
+ <edge from-layer="535" from-port="2" to-layer="537" to-port="0"/>
+ <edge from-layer="536" from-port="0" to-layer="537" to-port="1"/>
+ <edge from-layer="537" from-port="2" to-layer="539" to-port="0"/>
+ <edge from-layer="537" from-port="2" to-layer="554" to-port="0"/>
+ <edge from-layer="538" from-port="0" to-layer="539" to-port="1"/>
+ <edge from-layer="539" from-port="2" to-layer="541" to-port="0"/>
+ <edge from-layer="540" from-port="0" to-layer="541" to-port="1"/>
+ <edge from-layer="541" from-port="2" to-layer="543" to-port="0"/>
+ <edge from-layer="542" from-port="0" to-layer="543" to-port="1"/>
+ <edge from-layer="543" from-port="2" to-layer="545" to-port="0"/>
+ <edge from-layer="544" from-port="0" to-layer="545" to-port="1"/>
+ <edge from-layer="545" from-port="2" to-layer="547" to-port="0"/>
+ <edge from-layer="546" from-port="0" to-layer="547" to-port="1"/>
+ <edge from-layer="547" from-port="2" to-layer="549" to-port="0"/>
+ <edge from-layer="548" from-port="0" to-layer="549" to-port="1"/>
+ <edge from-layer="549" from-port="2" to-layer="551" to-port="0"/>
+ <edge from-layer="550" from-port="0" to-layer="551" to-port="1"/>
+ <edge from-layer="551" from-port="2" to-layer="553" to-port="0"/>
+ <edge from-layer="552" from-port="0" to-layer="553" to-port="1"/>
+ <edge from-layer="553" from-port="2" to-layer="554" to-port="1"/>
+ <edge from-layer="554" from-port="2" to-layer="556" to-port="0"/>
+ <edge from-layer="555" from-port="0" to-layer="556" to-port="1"/>
+ <edge from-layer="556" from-port="2" to-layer="558" to-port="0"/>
+ <edge from-layer="556" from-port="2" to-layer="573" to-port="0"/>
+ <edge from-layer="557" from-port="0" to-layer="558" to-port="1"/>
+ <edge from-layer="558" from-port="2" to-layer="560" to-port="0"/>
+ <edge from-layer="559" from-port="0" to-layer="560" to-port="1"/>
+ <edge from-layer="560" from-port="2" to-layer="562" to-port="0"/>
+ <edge from-layer="561" from-port="0" to-layer="562" to-port="1"/>
+ <edge from-layer="562" from-port="2" to-layer="564" to-port="0"/>
+ <edge from-layer="563" from-port="0" to-layer="564" to-port="1"/>
+ <edge from-layer="564" from-port="2" to-layer="566" to-port="0"/>
+ <edge from-layer="565" from-port="0" to-layer="566" to-port="1"/>
+ <edge from-layer="566" from-port="2" to-layer="568" to-port="0"/>
+ <edge from-layer="567" from-port="0" to-layer="568" to-port="1"/>
+ <edge from-layer="568" from-port="2" to-layer="570" to-port="0"/>
+ <edge from-layer="569" from-port="0" to-layer="570" to-port="1"/>
+ <edge from-layer="570" from-port="2" to-layer="572" to-port="0"/>
+ <edge from-layer="571" from-port="0" to-layer="572" to-port="1"/>
+ <edge from-layer="572" from-port="2" to-layer="573" to-port="1"/>
+ <edge from-layer="573" from-port="2" to-layer="575" to-port="0"/>
+ <edge from-layer="574" from-port="0" to-layer="575" to-port="1"/>
+ <edge from-layer="575" from-port="2" to-layer="592" to-port="0"/>
+ <edge from-layer="575" from-port="2" to-layer="577" to-port="0"/>
+ <edge from-layer="576" from-port="0" to-layer="577" to-port="1"/>
+ <edge from-layer="577" from-port="2" to-layer="579" to-port="0"/>
+ <edge from-layer="578" from-port="0" to-layer="579" to-port="1"/>
+ <edge from-layer="579" from-port="2" to-layer="581" to-port="0"/>
+ <edge from-layer="580" from-port="0" to-layer="581" to-port="1"/>
+ <edge from-layer="581" from-port="2" to-layer="583" to-port="0"/>
+ <edge from-layer="582" from-port="0" to-layer="583" to-port="1"/>
+ <edge from-layer="583" from-port="2" to-layer="585" to-port="0"/>
+ <edge from-layer="584" from-port="0" to-layer="585" to-port="1"/>
+ <edge from-layer="585" from-port="2" to-layer="587" to-port="0"/>
+ <edge from-layer="586" from-port="0" to-layer="587" to-port="1"/>
+ <edge from-layer="587" from-port="2" to-layer="589" to-port="0"/>
+ <edge from-layer="588" from-port="0" to-layer="589" to-port="1"/>
+ <edge from-layer="589" from-port="2" to-layer="591" to-port="0"/>
+ <edge from-layer="590" from-port="0" to-layer="591" to-port="1"/>
+ <edge from-layer="591" from-port="2" to-layer="592" to-port="1"/>
+ <edge from-layer="592" from-port="2" to-layer="594" to-port="0"/>
+ <edge from-layer="593" from-port="0" to-layer="594" to-port="1"/>
+ <edge from-layer="594" from-port="2" to-layer="611" to-port="0"/>
+ <edge from-layer="594" from-port="2" to-layer="596" to-port="0"/>
+ <edge from-layer="595" from-port="0" to-layer="596" to-port="1"/>
+ <edge from-layer="596" from-port="2" to-layer="598" to-port="0"/>
+ <edge from-layer="597" from-port="0" to-layer="598" to-port="1"/>
+ <edge from-layer="598" from-port="2" to-layer="600" to-port="0"/>
+ <edge from-layer="599" from-port="0" to-layer="600" to-port="1"/>
+ <edge from-layer="600" from-port="2" to-layer="602" to-port="0"/>
+ <edge from-layer="601" from-port="0" to-layer="602" to-port="1"/>
+ <edge from-layer="602" from-port="2" to-layer="604" to-port="0"/>
+ <edge from-layer="603" from-port="0" to-layer="604" to-port="1"/>
+ <edge from-layer="604" from-port="2" to-layer="606" to-port="0"/>
+ <edge from-layer="605" from-port="0" to-layer="606" to-port="1"/>
+ <edge from-layer="606" from-port="2" to-layer="608" to-port="0"/>
+ <edge from-layer="607" from-port="0" to-layer="608" to-port="1"/>
+ <edge from-layer="608" from-port="2" to-layer="610" to-port="0"/>
+ <edge from-layer="609" from-port="0" to-layer="610" to-port="1"/>
+ <edge from-layer="610" from-port="2" to-layer="611" to-port="1"/>
+ <edge from-layer="611" from-port="2" to-layer="613" to-port="0"/>
+ <edge from-layer="612" from-port="0" to-layer="613" to-port="1"/>
+ <edge from-layer="613" from-port="2" to-layer="615" to-port="0"/>
+ <edge from-layer="613" from-port="2" to-layer="630" to-port="0"/>
+ <edge from-layer="614" from-port="0" to-layer="615" to-port="1"/>
+ <edge from-layer="615" from-port="2" to-layer="617" to-port="0"/>
+ <edge from-layer="616" from-port="0" to-layer="617" to-port="1"/>
+ <edge from-layer="617" from-port="2" to-layer="619" to-port="0"/>
+ <edge from-layer="618" from-port="0" to-layer="619" to-port="1"/>
+ <edge from-layer="619" from-port="2" to-layer="621" to-port="0"/>
+ <edge from-layer="620" from-port="0" to-layer="621" to-port="1"/>
+ <edge from-layer="621" from-port="2" to-layer="623" to-port="0"/>
+ <edge from-layer="622" from-port="0" to-layer="623" to-port="1"/>
+ <edge from-layer="623" from-port="2" to-layer="625" to-port="0"/>
+ <edge from-layer="624" from-port="0" to-layer="625" to-port="1"/>
+ <edge from-layer="625" from-port="2" to-layer="627" to-port="0"/>
+ <edge from-layer="626" from-port="0" to-layer="627" to-port="1"/>
+ <edge from-layer="627" from-port="2" to-layer="629" to-port="0"/>
+ <edge from-layer="628" from-port="0" to-layer="629" to-port="1"/>
+ <edge from-layer="629" from-port="2" to-layer="630" to-port="1"/>
+ <edge from-layer="630" from-port="2" to-layer="632" to-port="0"/>
+ <edge from-layer="631" from-port="0" to-layer="632" to-port="1"/>
+ <edge from-layer="632" from-port="2" to-layer="649" to-port="0"/>
+ <edge from-layer="632" from-port="2" to-layer="634" to-port="0"/>
+ <edge from-layer="633" from-port="0" to-layer="634" to-port="1"/>
+ <edge from-layer="634" from-port="2" to-layer="636" to-port="0"/>
+ <edge from-layer="635" from-port="0" to-layer="636" to-port="1"/>
+ <edge from-layer="636" from-port="2" to-layer="638" to-port="0"/>
+ <edge from-layer="637" from-port="0" to-layer="638" to-port="1"/>
+ <edge from-layer="638" from-port="2" to-layer="640" to-port="0"/>
+ <edge from-layer="639" from-port="0" to-layer="640" to-port="1"/>
+ <edge from-layer="640" from-port="2" to-layer="642" to-port="0"/>
+ <edge from-layer="641" from-port="0" to-layer="642" to-port="1"/>
+ <edge from-layer="642" from-port="2" to-layer="644" to-port="0"/>
+ <edge from-layer="643" from-port="0" to-layer="644" to-port="1"/>
+ <edge from-layer="644" from-port="2" to-layer="646" to-port="0"/>
+ <edge from-layer="645" from-port="0" to-layer="646" to-port="1"/>
+ <edge from-layer="646" from-port="2" to-layer="648" to-port="0"/>
+ <edge from-layer="647" from-port="0" to-layer="648" to-port="1"/>
+ <edge from-layer="648" from-port="2" to-layer="649" to-port="1"/>
+ <edge from-layer="649" from-port="2" to-layer="651" to-port="0"/>
+ <edge from-layer="650" from-port="0" to-layer="651" to-port="1"/>
+ <edge from-layer="651" from-port="2" to-layer="653" to-port="0"/>
+ <edge from-layer="651" from-port="2" to-layer="668" to-port="0"/>
+ <edge from-layer="652" from-port="0" to-layer="653" to-port="1"/>
+ <edge from-layer="653" from-port="2" to-layer="655" to-port="0"/>
+ <edge from-layer="654" from-port="0" to-layer="655" to-port="1"/>
+ <edge from-layer="655" from-port="2" to-layer="657" to-port="0"/>
+ <edge from-layer="656" from-port="0" to-layer="657" to-port="1"/>
+ <edge from-layer="657" from-port="2" to-layer="659" to-port="0"/>
+ <edge from-layer="658" from-port="0" to-layer="659" to-port="1"/>
+ <edge from-layer="659" from-port="2" to-layer="661" to-port="0"/>
+ <edge from-layer="660" from-port="0" to-layer="661" to-port="1"/>
+ <edge from-layer="661" from-port="2" to-layer="663" to-port="0"/>
+ <edge from-layer="662" from-port="0" to-layer="663" to-port="1"/>
+ <edge from-layer="663" from-port="2" to-layer="665" to-port="0"/>
+ <edge from-layer="664" from-port="0" to-layer="665" to-port="1"/>
+ <edge from-layer="665" from-port="2" to-layer="667" to-port="0"/>
+ <edge from-layer="666" from-port="0" to-layer="667" to-port="1"/>
+ <edge from-layer="667" from-port="2" to-layer="668" to-port="1"/>
+ <edge from-layer="668" from-port="2" to-layer="670" to-port="0"/>
+ <edge from-layer="669" from-port="0" to-layer="670" to-port="1"/>
+ <edge from-layer="670" from-port="2" to-layer="672" to-port="0"/>
+ <edge from-layer="670" from-port="2" to-layer="687" to-port="0"/>
+ <edge from-layer="671" from-port="0" to-layer="672" to-port="1"/>
+ <edge from-layer="672" from-port="2" to-layer="674" to-port="0"/>
+ <edge from-layer="673" from-port="0" to-layer="674" to-port="1"/>
+ <edge from-layer="674" from-port="2" to-layer="676" to-port="0"/>
+ <edge from-layer="675" from-port="0" to-layer="676" to-port="1"/>
+ <edge from-layer="676" from-port="2" to-layer="678" to-port="0"/>
+ <edge from-layer="677" from-port="0" to-layer="678" to-port="1"/>
+ <edge from-layer="678" from-port="2" to-layer="680" to-port="0"/>
+ <edge from-layer="679" from-port="0" to-layer="680" to-port="1"/>
+ <edge from-layer="680" from-port="2" to-layer="682" to-port="0"/>
+ <edge from-layer="681" from-port="0" to-layer="682" to-port="1"/>
+ <edge from-layer="682" from-port="2" to-layer="684" to-port="0"/>
+ <edge from-layer="683" from-port="0" to-layer="684" to-port="1"/>
+ <edge from-layer="684" from-port="2" to-layer="686" to-port="0"/>
+ <edge from-layer="685" from-port="0" to-layer="686" to-port="1"/>
+ <edge from-layer="686" from-port="2" to-layer="687" to-port="1"/>
+ <edge from-layer="687" from-port="2" to-layer="689" to-port="0"/>
+ <edge from-layer="688" from-port="0" to-layer="689" to-port="1"/>
+ <edge from-layer="689" from-port="2" to-layer="706" to-port="0"/>
+ <edge from-layer="689" from-port="2" to-layer="691" to-port="0"/>
+ <edge from-layer="690" from-port="0" to-layer="691" to-port="1"/>
+ <edge from-layer="691" from-port="2" to-layer="693" to-port="0"/>
+ <edge from-layer="692" from-port="0" to-layer="693" to-port="1"/>
+ <edge from-layer="693" from-port="2" to-layer="695" to-port="0"/>
+ <edge from-layer="694" from-port="0" to-layer="695" to-port="1"/>
+ <edge from-layer="695" from-port="2" to-layer="697" to-port="0"/>
+ <edge from-layer="696" from-port="0" to-layer="697" to-port="1"/>
+ <edge from-layer="697" from-port="2" to-layer="699" to-port="0"/>
+ <edge from-layer="698" from-port="0" to-layer="699" to-port="1"/>
+ <edge from-layer="699" from-port="2" to-layer="701" to-port="0"/>
+ <edge from-layer="700" from-port="0" to-layer="701" to-port="1"/>
+ <edge from-layer="701" from-port="2" to-layer="703" to-port="0"/>
+ <edge from-layer="702" from-port="0" to-layer="703" to-port="1"/>
+ <edge from-layer="703" from-port="2" to-layer="705" to-port="0"/>
+ <edge from-layer="704" from-port="0" to-layer="705" to-port="1"/>
+ <edge from-layer="705" from-port="2" to-layer="706" to-port="1"/>
+ <edge from-layer="706" from-port="2" to-layer="708" to-port="0"/>
+ <edge from-layer="707" from-port="0" to-layer="708" to-port="1"/>
+ <edge from-layer="708" from-port="2" to-layer="730" to-port="0"/>
+ <edge from-layer="708" from-port="2" to-layer="718" to-port="0"/>
+ <edge from-layer="708" from-port="2" to-layer="710" to-port="0"/>
+ <edge from-layer="709" from-port="0" to-layer="710" to-port="1"/>
+ <edge from-layer="710" from-port="2" to-layer="712" to-port="0"/>
+ <edge from-layer="711" from-port="0" to-layer="712" to-port="1"/>
+ <edge from-layer="712" from-port="2" to-layer="714" to-port="0"/>
+ <edge from-layer="713" from-port="0" to-layer="714" to-port="1"/>
+ <edge from-layer="714" from-port="2" to-layer="716" to-port="0"/>
+ <edge from-layer="715" from-port="0" to-layer="716" to-port="1"/>
+ <edge from-layer="716" from-port="2" to-layer="743" to-port="0"/>
+ <edge from-layer="717" from-port="0" to-layer="718" to-port="1"/>
+ <edge from-layer="718" from-port="2" to-layer="720" to-port="0"/>
+ <edge from-layer="719" from-port="0" to-layer="720" to-port="1"/>
+ <edge from-layer="720" from-port="2" to-layer="722" to-port="0"/>
+ <edge from-layer="721" from-port="0" to-layer="722" to-port="1"/>
+ <edge from-layer="722" from-port="2" to-layer="724" to-port="0"/>
+ <edge from-layer="723" from-port="0" to-layer="724" to-port="1"/>
+ <edge from-layer="724" from-port="2" to-layer="726" to-port="0"/>
+ <edge from-layer="725" from-port="0" to-layer="726" to-port="1"/>
+ <edge from-layer="726" from-port="2" to-layer="727" to-port="0"/>
+ <edge from-layer="727" from-port="1" to-layer="729" to-port="0"/>
+ <edge from-layer="728" from-port="0" to-layer="729" to-port="1"/>
+ <edge from-layer="729" from-port="2" to-layer="743" to-port="1"/>
+ <edge from-layer="730" from-port="1" to-layer="734" to-port="0"/>
+ <edge from-layer="731" from-port="0" to-layer="734" to-port="1"/>
+ <edge from-layer="732" from-port="0" to-layer="734" to-port="2"/>
+ <edge from-layer="733" from-port="0" to-layer="734" to-port="3"/>
+ <edge from-layer="734" from-port="4" to-layer="740" to-port="0"/>
+ <edge from-layer="735" from-port="1" to-layer="739" to-port="0"/>
+ <edge from-layer="736" from-port="0" to-layer="739" to-port="1"/>
+ <edge from-layer="737" from-port="0" to-layer="739" to-port="2"/>
+ <edge from-layer="738" from-port="0" to-layer="739" to-port="3"/>
+ <edge from-layer="739" from-port="4" to-layer="740" to-port="1"/>
+ <edge from-layer="740" from-port="2" to-layer="742" to-port="0"/>
+ <edge from-layer="741" from-port="0" to-layer="742" to-port="1"/>
+ <edge from-layer="742" from-port="2" to-layer="743" to-port="2"/>
+ <edge from-layer="743" from-port="3" to-layer="744" to-port="0"/>
+ </edges>
+ <meta_data>
+ <MO_version value="2021.4.0-3827-c5b65f2cb1d-releases/2021/4"/>
+ <cli_parameters>
+ <caffe_parser_path value="DIR"/>
+ <data_type value="FP16"/>
+ <disable_nhwc_to_nchw value="False"/>
+ <disable_omitting_optional value="False"/>
+ <disable_resnet_optimization value="False"/>
+ <disable_weights_compression value="False"/>
+ <enable_concat_optimization value="False"/>
+ <enable_flattening_nested_params value="False"/>
+ <enable_ssd_gluoncv value="False"/>
+ <extensions value="DIR"/>
+ <framework value="caffe"/>
+ <freeze_placeholder_with_value value="{}"/>
+ <generate_deprecated_IR_V7 value="False"/>
+ <input value="data"/>
+ <input_model value="DIR/rmnet_lrelu_pd_ssd.caffemodel"/>
+ <input_model_is_text value="False"/>
+ <input_proto value="DIR/rmnet_lrelu_pd_ssd.prototxt"/>
+ <input_shape value="[1,3,320,544]"/>
+ <k value="DIR/CustomLayersMapping.xml"/>
+ <keep_shape_ops value="True"/>
+ <legacy_ir_generation value="False"/>
+ <legacy_mxnet_model value="False"/>
+ <log_level value="ERROR"/>
+ <mean_scale_values value="{}"/>
+ <mean_values value="()"/>
+ <model_name value="person-detection-retail-0013"/>
+ <output value="['detection_out']"/>
+ <output_dir value="DIR"/>
+ <placeholder_data_types value="{}"/>
+ <placeholder_shapes value="{'data': array([ 1, 3, 320, 544])}"/>
+ <progress value="False"/>
+ <remove_memory value="False"/>
+ <remove_output_softmax value="False"/>
+ <reverse_input_channels value="False"/>
+ <save_params_from_nd value="False"/>
+ <scale_values value="()"/>
+ <silent value="False"/>
+ <static_shape value="False"/>
+ <stream_output value="False"/>
+ <transform value=""/>
+ <unset unset_cli_parameters="batch, counts, disable_fusing, disable_gfusing, finegrain_fusing, input_checkpoint, input_meta_graph, input_symbol, mean_file, mean_file_offsets, move_to_preprocess, nd_prefix_name, pretrained_model_name, saved_model_dir, saved_model_tags, scale, tensorboard_logdir, tensorflow_custom_layer_libraries, tensorflow_custom_operations_config_update, tensorflow_object_detection_api_pipeline_config, tensorflow_use_custom_operations_config, transformations_config"/>
+ </cli_parameters>
+ </meta_data>
+</net>
diff --git a/resources/models/intel/person-detection-retail-0013/FP32/person-detection-retail-0013.bin b/resources/models/intel/person-detection-retail-0013/FP32/person-detection-retail-0013.bin
new file mode 100644
index 0000000..4719f58
--- /dev/null
+++ b/resources/models/intel/person-detection-retail-0013/FP32/person-detection-retail-0013.bin
Binary files differ
diff --git a/resources/models/intel/person-detection-retail-0013/FP32/person-detection-retail-0013.xml b/resources/models/intel/person-detection-retail-0013/FP32/person-detection-retail-0013.xml
new file mode 100644
index 0000000..8be2b9f
--- /dev/null
+++ b/resources/models/intel/person-detection-retail-0013/FP32/person-detection-retail-0013.xml
@@ -0,0 +1,13694 @@
+<?xml version="1.0" ?>
+<net name="ResMobNet_v4 (LReLU) with single SSD head" version="10">
+ <layers>
+ <layer id="0" name="data" type="Parameter" version="opset1">
+ <data element_type="f32" shape="1, 3, 320, 544"/>
+ <output>
+ <port id="0" names="data" precision="FP32">
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>320</dim>
+ <dim>544</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="1" name="data_mul_23644" type="Const" version="opset1">
+ <data element_type="f32" offset="0" shape="1, 3, 1, 1" size="12"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="2" name="data/norm/bn/mean/Fused_Mul_" type="Multiply" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>320</dim>
+ <dim>544</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP32">
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>320</dim>
+ <dim>544</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="3" name="data_add_23646" type="Const" version="opset1">
+ <data element_type="f32" offset="12" shape="1, 3, 1, 1" size="12"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="4" name="data/norm/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>320</dim>
+ <dim>544</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="data/norm/bn" precision="FP32">
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>320</dim>
+ <dim>544</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="5" name="init_block1/dim_inc/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f32" offset="24" shape="32, 3, 3, 3" size="3456"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>32</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="6" name="init_block1/dim_inc/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="1, 1" pads_end="1, 1" strides="2, 2"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>320</dim>
+ <dim>544</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>32</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="7" name="data_add_2364923654" type="Const" version="opset1">
+ <data element_type="f32" offset="3480" shape="1, 32, 1, 1" size="128"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="8" name="init_block1/dim_inc/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="init_block1/dim_inc/conv" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="9" name="init_block1/dim_inc/fn" type="ReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ </input>
+ <output>
+ <port id="1" names="init_block1/dim_inc/conv" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="10" name="bottleneck1_1/dim_red/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f32" offset="3608" shape="8, 32, 1, 1" size="1024"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>8</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="11" name="bottleneck1_1/dim_red/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="0, 0" pads_end="0, 0" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>8</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP32">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="12" name="data_add_2365723662" type="Const" version="opset1">
+ <data element_type="f32" offset="4632" shape="1, 8, 1, 1" size="32"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="13" name="bottleneck1_1/dim_red/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck1_1/dim_red/conv" precision="FP32">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="14" name="bottleneck1_1/dim_red/fn/weights3096039785" type="Const" version="opset1">
+ <data element_type="f32" offset="4664" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="15" name="bottleneck1_1/dim_red/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck1_1/dim_red/conv" precision="FP32">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="16" name="bottleneck1_1/inner/dw1/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f32" offset="4668" shape="8, 1, 1, 3, 3" size="288"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="17" name="bottleneck1_1/inner/dw1/conv" type="GroupConvolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="1, 1" pads_end="1, 1" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP32">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="18" name="data_add_2366523670" type="Const" version="opset1">
+ <data element_type="f32" offset="4956" shape="1, 8, 1, 1" size="32"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="19" name="bottleneck1_1/inner/dw1/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck1_1/inner/dw1/conv" precision="FP32">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="20" name="bottleneck1_1/inner/dw1/fn/weights3102439659" type="Const" version="opset1">
+ <data element_type="f32" offset="4664" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="21" name="bottleneck1_1/inner/dw1/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck1_1/inner/dw1/conv" precision="FP32">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="22" name="bottleneck1_1/dim_inc/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f32" offset="4988" shape="32, 8, 1, 1" size="1024"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>32</dim>
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="23" name="bottleneck1_1/dim_inc/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="0, 0" pads_end="0, 0" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>32</dim>
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="24" name="data_add_2367323678" type="Const" version="opset1">
+ <data element_type="f32" offset="6012" shape="1, 32, 1, 1" size="128"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="25" name="bottleneck1_1/dim_inc/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck1_1/dim_inc/conv" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="26" name="bottleneck1_1/add" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck1_1/add" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="27" name="bottleneck1_1/fn/weights3115239677" type="Const" version="opset1">
+ <data element_type="f32" offset="4664" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="28" name="bottleneck1_1/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck1_1/add" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="29" name="bottleneck1_2/dim_red/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f32" offset="6140" shape="8, 32, 1, 1" size="1024"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>8</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="30" name="bottleneck1_2/dim_red/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="0, 0" pads_end="0, 0" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>8</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP32">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="31" name="data_add_2368123686" type="Const" version="opset1">
+ <data element_type="f32" offset="7164" shape="1, 8, 1, 1" size="32"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="32" name="bottleneck1_2/dim_red/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck1_2/dim_red/conv" precision="FP32">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="33" name="bottleneck1_2/dim_red/fn/weights3077639878" type="Const" version="opset1">
+ <data element_type="f32" offset="4664" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="34" name="bottleneck1_2/dim_red/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck1_2/dim_red/conv" precision="FP32">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="35" name="bottleneck1_2/inner/dw1/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f32" offset="7196" shape="8, 1, 1, 3, 3" size="288"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="36" name="bottleneck1_2/inner/dw1/conv" type="GroupConvolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="1, 1" pads_end="1, 1" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP32">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="37" name="data_add_2368923694" type="Const" version="opset1">
+ <data element_type="f32" offset="7484" shape="1, 8, 1, 1" size="32"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="38" name="bottleneck1_2/inner/dw1/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck1_2/inner/dw1/conv" precision="FP32">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="39" name="bottleneck1_2/inner/dw1/fn/weights3087240085" type="Const" version="opset1">
+ <data element_type="f32" offset="4664" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="40" name="bottleneck1_2/inner/dw1/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck1_2/inner/dw1/conv" precision="FP32">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="41" name="bottleneck1_2/dim_inc/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f32" offset="7516" shape="32, 8, 1, 1" size="1024"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>32</dim>
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="42" name="bottleneck1_2/dim_inc/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="0, 0" pads_end="0, 0" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>32</dim>
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="43" name="data_add_2369723702" type="Const" version="opset1">
+ <data element_type="f32" offset="8540" shape="1, 32, 1, 1" size="128"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="44" name="bottleneck1_2/dim_inc/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck1_2/dim_inc/conv" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="45" name="bottleneck1_2/add" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck1_2/add" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="46" name="bottleneck1_2/fn/weights3090439737" type="Const" version="opset1">
+ <data element_type="f32" offset="4664" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="47" name="bottleneck1_2/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck1_2/add" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="48" name="bottleneck1_3/dim_red/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f32" offset="8668" shape="8, 32, 1, 1" size="1024"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>8</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="49" name="bottleneck1_3/dim_red/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="0, 0" pads_end="0, 0" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>8</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP32">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="50" name="data_add_2370523710" type="Const" version="opset1">
+ <data element_type="f32" offset="9692" shape="1, 8, 1, 1" size="32"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="51" name="bottleneck1_3/dim_red/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck1_3/dim_red/conv" precision="FP32">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="52" name="bottleneck1_3/dim_red/fn/weights3092840502" type="Const" version="opset1">
+ <data element_type="f32" offset="4664" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="53" name="bottleneck1_3/dim_red/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck1_3/dim_red/conv" precision="FP32">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="54" name="bottleneck1_3/inner/dw1/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f32" offset="9724" shape="8, 1, 1, 3, 3" size="288"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="55" name="bottleneck1_3/inner/dw1/conv" type="GroupConvolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="1, 1" pads_end="1, 1" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP32">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="56" name="data_add_2371323718" type="Const" version="opset1">
+ <data element_type="f32" offset="10012" shape="1, 8, 1, 1" size="32"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="57" name="bottleneck1_3/inner/dw1/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck1_3/inner/dw1/conv" precision="FP32">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="58" name="bottleneck1_3/inner/dw1/fn/weights3115640004" type="Const" version="opset1">
+ <data element_type="f32" offset="4664" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="59" name="bottleneck1_3/inner/dw1/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck1_3/inner/dw1/conv" precision="FP32">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="60" name="bottleneck1_3/dim_inc/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f32" offset="10044" shape="32, 8, 1, 1" size="1024"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>32</dim>
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="61" name="bottleneck1_3/dim_inc/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="0, 0" pads_end="0, 0" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>32</dim>
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="62" name="data_add_2372123726" type="Const" version="opset1">
+ <data element_type="f32" offset="11068" shape="1, 32, 1, 1" size="128"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="63" name="bottleneck1_3/dim_inc/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck1_3/dim_inc/conv" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="64" name="bottleneck1_3/add" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck1_3/add" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="65" name="bottleneck1_3/fn/weights3092439836" type="Const" version="opset1">
+ <data element_type="f32" offset="4664" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="66" name="bottleneck1_3/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck1_3/add" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="67" name="bottleneck1_4/dim_red/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f32" offset="11196" shape="8, 32, 1, 1" size="1024"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>8</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="68" name="bottleneck1_4/dim_red/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="0, 0" pads_end="0, 0" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>8</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP32">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="69" name="data_add_2372923734" type="Const" version="opset1">
+ <data element_type="f32" offset="12220" shape="1, 8, 1, 1" size="32"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="70" name="bottleneck1_4/dim_red/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck1_4/dim_red/conv" precision="FP32">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="71" name="bottleneck1_4/dim_red/fn/weights3114840622" type="Const" version="opset1">
+ <data element_type="f32" offset="4664" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="72" name="bottleneck1_4/dim_red/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck1_4/dim_red/conv" precision="FP32">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="73" name="bottleneck1_4/inner/dw1/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f32" offset="12252" shape="8, 1, 1, 3, 3" size="288"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="74" name="bottleneck1_4/inner/dw1/conv" type="GroupConvolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="1, 1" pads_end="1, 1" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP32">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="75" name="data_add_2373723742" type="Const" version="opset1">
+ <data element_type="f32" offset="12540" shape="1, 8, 1, 1" size="32"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="76" name="bottleneck1_4/inner/dw1/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck1_4/inner/dw1/conv" precision="FP32">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="77" name="bottleneck1_4/inner/dw1/fn/weights3095640181" type="Const" version="opset1">
+ <data element_type="f32" offset="4664" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="78" name="bottleneck1_4/inner/dw1/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck1_4/inner/dw1/conv" precision="FP32">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="79" name="bottleneck1_4/dim_inc/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f32" offset="12572" shape="32, 8, 1, 1" size="1024"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>32</dim>
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="80" name="bottleneck1_4/dim_inc/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="0, 0" pads_end="0, 0" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>8</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>32</dim>
+ <dim>8</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="81" name="data_add_2374523750" type="Const" version="opset1">
+ <data element_type="f32" offset="13596" shape="1, 32, 1, 1" size="128"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="82" name="bottleneck1_4/dim_inc/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck1_4/dim_inc/conv" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="83" name="bottleneck1_4/add" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck1_4/add" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="84" name="bottleneck1_4/fn/weights3087639962" type="Const" version="opset1">
+ <data element_type="f32" offset="4664" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="85" name="bottleneck1_4/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck1_4/add" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="86" name="bottleneck2_0/skip/pooling" type="MaxPool" version="opset1">
+ <data auto_pad="explicit" kernel="2, 2" pads_begin="0, 0" pads_end="0, 0" rounding_type="ceil" strides="2, 2"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ </input>
+ <output>
+ <port id="1" names="bottleneck2_0/skip/pooling" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="87" name="bottleneck2_0/skip/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f32" offset="13724" shape="64, 32, 1, 1" size="8192"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>64</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="88" name="bottleneck2_0/skip/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="0, 0" pads_end="0, 0" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>64</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="89" name="data_add_2375323758" type="Const" version="opset1">
+ <data element_type="f32" offset="21916" shape="1, 64, 1, 1" size="256"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="90" name="bottleneck2_0/skip/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck2_0/skip/conv" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="91" name="bottleneck2_0/dim_red/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f32" offset="22172" shape="16, 32, 1, 1" size="2048"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>16</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="92" name="bottleneck2_0/dim_red/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="0, 0" pads_end="0, 0" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>16</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP32">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="93" name="data_add_2376123766" type="Const" version="opset1">
+ <data element_type="f32" offset="24220" shape="1, 16, 1, 1" size="64"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="94" name="bottleneck2_0/dim_red/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck2_0/dim_red/conv" precision="FP32">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="95" name="bottleneck2_0/dim_red/fn/weights3103239749" type="Const" version="opset1">
+ <data element_type="f32" offset="4664" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="96" name="bottleneck2_0/dim_red/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck2_0/dim_red/conv" precision="FP32">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="97" name="bottleneck2_0/inner/dw1/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f32" offset="24284" shape="16, 1, 1, 3, 3" size="576"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="98" name="bottleneck2_0/inner/dw1/conv" type="GroupConvolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="1, 1" pads_end="1, 1" strides="2, 2"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>160</dim>
+ <dim>272</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP32">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="99" name="data_add_2376923774" type="Const" version="opset1">
+ <data element_type="f32" offset="24860" shape="1, 16, 1, 1" size="64"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="100" name="bottleneck2_0/inner/dw1/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck2_0/inner/dw1/conv" precision="FP32">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="101" name="bottleneck2_0/inner/dw1/fn/weights3088840568" type="Const" version="opset1">
+ <data element_type="f32" offset="4664" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="102" name="bottleneck2_0/inner/dw1/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck2_0/inner/dw1/conv" precision="FP32">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="103" name="bottleneck2_0/dim_inc/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f32" offset="24924" shape="64, 16, 1, 1" size="4096"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>64</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="104" name="bottleneck2_0/dim_inc/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="0, 0" pads_end="0, 0" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>64</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="105" name="data_add_2377723782" type="Const" version="opset1">
+ <data element_type="f32" offset="29020" shape="1, 64, 1, 1" size="256"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="106" name="bottleneck2_0/dim_inc/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck2_0/dim_inc/conv" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="107" name="bottleneck2_0/add" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck2_0/add" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="108" name="bottleneck2_0/fn/weights3086440226" type="Const" version="opset1">
+ <data element_type="f32" offset="4664" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="109" name="bottleneck2_0/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck2_0/add" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="110" name="bottleneck2_1/dim_red/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f32" offset="29276" shape="16, 64, 1, 1" size="4096"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>16</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="111" name="bottleneck2_1/dim_red/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="0, 0" pads_end="0, 0" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>16</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP32">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="112" name="data_add_2378523790" type="Const" version="opset1">
+ <data element_type="f32" offset="33372" shape="1, 16, 1, 1" size="64"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="113" name="bottleneck2_1/dim_red/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck2_1/dim_red/conv" precision="FP32">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="114" name="bottleneck2_1/dim_red/fn/weights3091240172" type="Const" version="opset1">
+ <data element_type="f32" offset="4664" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="115" name="bottleneck2_1/dim_red/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck2_1/dim_red/conv" precision="FP32">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="116" name="bottleneck2_1/inner/dw1/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f32" offset="33436" shape="16, 1, 1, 3, 3" size="576"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="117" name="bottleneck2_1/inner/dw1/conv" type="GroupConvolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="1, 1" pads_end="1, 1" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP32">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="118" name="data_add_2379323798" type="Const" version="opset1">
+ <data element_type="f32" offset="34012" shape="1, 16, 1, 1" size="64"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="119" name="bottleneck2_1/inner/dw1/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck2_1/inner/dw1/conv" precision="FP32">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="120" name="bottleneck2_1/inner/dw1/fn/weights3110039803" type="Const" version="opset1">
+ <data element_type="f32" offset="4664" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="121" name="bottleneck2_1/inner/dw1/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck2_1/inner/dw1/conv" precision="FP32">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="122" name="bottleneck2_1/dim_inc/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f32" offset="34076" shape="64, 16, 1, 1" size="4096"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>64</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="123" name="bottleneck2_1/dim_inc/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="0, 0" pads_end="0, 0" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>64</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="124" name="data_add_2380123806" type="Const" version="opset1">
+ <data element_type="f32" offset="38172" shape="1, 64, 1, 1" size="256"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="125" name="bottleneck2_1/dim_inc/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck2_1/dim_inc/conv" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="126" name="bottleneck2_1/add" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck2_1/add" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="127" name="bottleneck2_1/fn/weights3081640076" type="Const" version="opset1">
+ <data element_type="f32" offset="4664" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="128" name="bottleneck2_1/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck2_1/add" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="129" name="bottleneck2_2/dim_red/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f32" offset="38428" shape="16, 64, 1, 1" size="4096"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>16</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="130" name="bottleneck2_2/dim_red/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="0, 0" pads_end="0, 0" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>16</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP32">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="131" name="data_add_2380923814" type="Const" version="opset1">
+ <data element_type="f32" offset="42524" shape="1, 16, 1, 1" size="64"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="132" name="bottleneck2_2/dim_red/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck2_2/dim_red/conv" precision="FP32">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="133" name="bottleneck2_2/dim_red/fn/weights3079239824" type="Const" version="opset1">
+ <data element_type="f32" offset="4664" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="134" name="bottleneck2_2/dim_red/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck2_2/dim_red/conv" precision="FP32">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="135" name="bottleneck2_2/inner/dw1/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f32" offset="42588" shape="16, 1, 1, 3, 3" size="576"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="136" name="bottleneck2_2/inner/dw1/conv" type="GroupConvolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="1, 1" pads_end="1, 1" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP32">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="137" name="data_add_2381723822" type="Const" version="opset1">
+ <data element_type="f32" offset="43164" shape="1, 16, 1, 1" size="64"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="138" name="bottleneck2_2/inner/dw1/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck2_2/inner/dw1/conv" precision="FP32">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="139" name="bottleneck2_2/inner/dw1/fn/weights3110439791" type="Const" version="opset1">
+ <data element_type="f32" offset="4664" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="140" name="bottleneck2_2/inner/dw1/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck2_2/inner/dw1/conv" precision="FP32">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="141" name="bottleneck2_2/dim_inc/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f32" offset="43228" shape="64, 16, 1, 1" size="4096"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>64</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="142" name="bottleneck2_2/dim_inc/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="0, 0" pads_end="0, 0" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>64</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="143" name="data_add_2382523830" type="Const" version="opset1">
+ <data element_type="f32" offset="47324" shape="1, 64, 1, 1" size="256"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="144" name="bottleneck2_2/dim_inc/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck2_2/dim_inc/conv" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="145" name="bottleneck2_2/add" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck2_2/add" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="146" name="bottleneck2_2/fn/weights3100439671" type="Const" version="opset1">
+ <data element_type="f32" offset="4664" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="147" name="bottleneck2_2/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck2_2/add" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="148" name="bottleneck2_3/dim_red/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f32" offset="47580" shape="16, 64, 1, 1" size="4096"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>16</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="149" name="bottleneck2_3/dim_red/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="0, 0" pads_end="0, 0" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>16</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP32">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="150" name="data_add_2383323838" type="Const" version="opset1">
+ <data element_type="f32" offset="51676" shape="1, 16, 1, 1" size="64"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="151" name="bottleneck2_3/dim_red/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck2_3/dim_red/conv" precision="FP32">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="152" name="bottleneck2_3/dim_red/fn/weights3096440040" type="Const" version="opset1">
+ <data element_type="f32" offset="4664" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="153" name="bottleneck2_3/dim_red/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck2_3/dim_red/conv" precision="FP32">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="154" name="bottleneck2_3/inner/dw1/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f32" offset="51740" shape="16, 1, 1, 3, 3" size="576"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="155" name="bottleneck2_3/inner/dw1/conv" type="GroupConvolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="1, 1" pads_end="1, 1" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP32">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="156" name="data_add_2384123846" type="Const" version="opset1">
+ <data element_type="f32" offset="52316" shape="1, 16, 1, 1" size="64"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="157" name="bottleneck2_3/inner/dw1/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck2_3/inner/dw1/conv" precision="FP32">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="158" name="bottleneck2_3/inner/dw1/fn/weights3080039752" type="Const" version="opset1">
+ <data element_type="f32" offset="4664" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="159" name="bottleneck2_3/inner/dw1/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck2_3/inner/dw1/conv" precision="FP32">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="160" name="bottleneck2_3/dim_inc/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f32" offset="52380" shape="64, 16, 1, 1" size="4096"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>64</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="161" name="bottleneck2_3/dim_inc/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="0, 0" pads_end="0, 0" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>64</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="162" name="data_add_2384923854" type="Const" version="opset1">
+ <data element_type="f32" offset="56476" shape="1, 64, 1, 1" size="256"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="163" name="bottleneck2_3/dim_inc/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck2_3/dim_inc/conv" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="164" name="bottleneck2_3/add" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck2_3/add" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="165" name="bottleneck2_3/fn/weights3076840016" type="Const" version="opset1">
+ <data element_type="f32" offset="4664" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="166" name="bottleneck2_3/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck2_3/add" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="167" name="bottleneck2_4/dim_red/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f32" offset="56732" shape="16, 64, 1, 1" size="4096"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>16</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="168" name="bottleneck2_4/dim_red/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="0, 0" pads_end="0, 0" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>16</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP32">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="169" name="data_add_2385723862" type="Const" version="opset1">
+ <data element_type="f32" offset="60828" shape="1, 16, 1, 1" size="64"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="170" name="bottleneck2_4/dim_red/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck2_4/dim_red/conv" precision="FP32">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="171" name="bottleneck2_4/dim_red/fn/weights3085640454" type="Const" version="opset1">
+ <data element_type="f32" offset="4664" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="172" name="bottleneck2_4/dim_red/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck2_4/dim_red/conv" precision="FP32">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="173" name="bottleneck2_4/inner/dw1/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f32" offset="60892" shape="16, 1, 1, 3, 3" size="576"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="174" name="bottleneck2_4/inner/dw1/conv" type="GroupConvolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="1, 1" pads_end="1, 1" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP32">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="175" name="data_add_2386523870" type="Const" version="opset1">
+ <data element_type="f32" offset="61468" shape="1, 16, 1, 1" size="64"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="176" name="bottleneck2_4/inner/dw1/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck2_4/inner/dw1/conv" precision="FP32">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="177" name="bottleneck2_4/inner/dw1/fn/weights3082039965" type="Const" version="opset1">
+ <data element_type="f32" offset="4664" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="178" name="bottleneck2_4/inner/dw1/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck2_4/inner/dw1/conv" precision="FP32">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="179" name="bottleneck2_4/dim_inc/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f32" offset="61532" shape="64, 16, 1, 1" size="4096"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>64</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="180" name="bottleneck2_4/dim_inc/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="0, 0" pads_end="0, 0" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>64</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="181" name="data_add_2387323878" type="Const" version="opset1">
+ <data element_type="f32" offset="65628" shape="1, 64, 1, 1" size="256"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="182" name="bottleneck2_4/dim_inc/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck2_4/dim_inc/conv" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="183" name="bottleneck2_4/add" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck2_4/add" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="184" name="bottleneck2_4/fn/weights3106840703" type="Const" version="opset1">
+ <data element_type="f32" offset="4664" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="185" name="bottleneck2_4/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck2_4/add" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="186" name="bottleneck2_5/dim_red/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f32" offset="65884" shape="16, 64, 1, 1" size="4096"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>16</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="187" name="bottleneck2_5/dim_red/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="0, 0" pads_end="0, 0" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>16</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP32">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="188" name="data_add_2388123886" type="Const" version="opset1">
+ <data element_type="f32" offset="69980" shape="1, 16, 1, 1" size="64"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="189" name="bottleneck2_5/dim_red/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck2_5/dim_red/conv" precision="FP32">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="190" name="bottleneck2_5/dim_red/fn/weights3082839890" type="Const" version="opset1">
+ <data element_type="f32" offset="4664" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="191" name="bottleneck2_5/dim_red/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck2_5/dim_red/conv" precision="FP32">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="192" name="bottleneck2_5/inner/dw1/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f32" offset="70044" shape="16, 1, 1, 3, 3" size="576"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="193" name="bottleneck2_5/inner/dw1/conv" type="GroupConvolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="1, 1" pads_end="1, 1" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP32">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="194" name="data_add_2388923894" type="Const" version="opset1">
+ <data element_type="f32" offset="70620" shape="1, 16, 1, 1" size="64"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="195" name="bottleneck2_5/inner/dw1/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck2_5/inner/dw1/conv" precision="FP32">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="196" name="bottleneck2_5/inner/dw1/fn/weights3088039707" type="Const" version="opset1">
+ <data element_type="f32" offset="4664" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="197" name="bottleneck2_5/inner/dw1/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck2_5/inner/dw1/conv" precision="FP32">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="198" name="bottleneck2_5/dim_inc/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f32" offset="70684" shape="64, 16, 1, 1" size="4096"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>64</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="199" name="bottleneck2_5/dim_inc/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="0, 0" pads_end="0, 0" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>64</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="200" name="data_add_2389723902" type="Const" version="opset1">
+ <data element_type="f32" offset="74780" shape="1, 64, 1, 1" size="256"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="201" name="bottleneck2_5/dim_inc/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck2_5/dim_inc/conv" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="202" name="bottleneck2_5/add" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck2_5/add" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="203" name="bottleneck2_5/fn/weights3085239683" type="Const" version="opset1">
+ <data element_type="f32" offset="4664" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="204" name="bottleneck2_5/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck2_5/add" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="205" name="bottleneck2_6/dim_red/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f32" offset="75036" shape="16, 64, 1, 1" size="4096"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>16</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="206" name="bottleneck2_6/dim_red/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="0, 0" pads_end="0, 0" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>16</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP32">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="207" name="data_add_2390523910" type="Const" version="opset1">
+ <data element_type="f32" offset="79132" shape="1, 16, 1, 1" size="64"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="208" name="bottleneck2_6/dim_red/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck2_6/dim_red/conv" precision="FP32">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="209" name="bottleneck2_6/dim_red/fn/weights3097240064" type="Const" version="opset1">
+ <data element_type="f32" offset="4664" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="210" name="bottleneck2_6/dim_red/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck2_6/dim_red/conv" precision="FP32">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="211" name="bottleneck2_6/inner/dw1/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f32" offset="79196" shape="16, 1, 1, 3, 3" size="576"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="212" name="bottleneck2_6/inner/dw1/conv" type="GroupConvolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="1, 1" pads_end="1, 1" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP32">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="213" name="data_add_2391323918" type="Const" version="opset1">
+ <data element_type="f32" offset="79772" shape="1, 16, 1, 1" size="64"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="214" name="bottleneck2_6/inner/dw1/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck2_6/inner/dw1/conv" precision="FP32">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="215" name="bottleneck2_6/inner/dw1/fn/weights3114439989" type="Const" version="opset1">
+ <data element_type="f32" offset="4664" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="216" name="bottleneck2_6/inner/dw1/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck2_6/inner/dw1/conv" precision="FP32">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="217" name="bottleneck2_6/dim_inc/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f32" offset="79836" shape="64, 16, 1, 1" size="4096"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>64</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="218" name="bottleneck2_6/dim_inc/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="0, 0" pads_end="0, 0" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>64</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="219" name="data_add_2392123926" type="Const" version="opset1">
+ <data element_type="f32" offset="83932" shape="1, 64, 1, 1" size="256"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="220" name="bottleneck2_6/dim_inc/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck2_6/dim_inc/conv" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="221" name="bottleneck2_6/add" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck2_6/add" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="222" name="bottleneck2_6/fn/weights3107640616" type="Const" version="opset1">
+ <data element_type="f32" offset="4664" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="223" name="bottleneck2_6/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck2_6/add" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="224" name="bottleneck2_7/dim_red/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f32" offset="84188" shape="16, 64, 1, 1" size="4096"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>16</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="225" name="bottleneck2_7/dim_red/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="0, 0" pads_end="0, 0" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>16</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP32">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="226" name="data_add_2392923934" type="Const" version="opset1">
+ <data element_type="f32" offset="88284" shape="1, 16, 1, 1" size="64"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="227" name="bottleneck2_7/dim_red/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck2_7/dim_red/conv" precision="FP32">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="228" name="bottleneck2_7/dim_red/fn/weights3110839845" type="Const" version="opset1">
+ <data element_type="f32" offset="4664" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="229" name="bottleneck2_7/dim_red/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck2_7/dim_red/conv" precision="FP32">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="230" name="bottleneck2_7/inner/dw1/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f32" offset="88348" shape="16, 1, 1, 3, 3" size="576"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="231" name="bottleneck2_7/inner/dw1/conv" type="GroupConvolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="1, 1" pads_end="1, 1" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP32">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="232" name="data_add_2393723942" type="Const" version="opset1">
+ <data element_type="f32" offset="88924" shape="1, 16, 1, 1" size="64"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="233" name="bottleneck2_7/inner/dw1/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck2_7/inner/dw1/conv" precision="FP32">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="234" name="bottleneck2_7/inner/dw1/fn/weights3118040055" type="Const" version="opset1">
+ <data element_type="f32" offset="4664" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="235" name="bottleneck2_7/inner/dw1/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck2_7/inner/dw1/conv" precision="FP32">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="236" name="bottleneck2_7/dim_inc/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f32" offset="88988" shape="64, 16, 1, 1" size="4096"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>64</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="237" name="bottleneck2_7/dim_inc/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="0, 0" pads_end="0, 0" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>64</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="238" name="data_add_2394523950" type="Const" version="opset1">
+ <data element_type="f32" offset="93084" shape="1, 64, 1, 1" size="256"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="239" name="bottleneck2_7/dim_inc/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck2_7/dim_inc/conv" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="240" name="bottleneck2_7/add" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck2_7/add" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="241" name="bottleneck2_7/fn/weights3106040265" type="Const" version="opset1">
+ <data element_type="f32" offset="4664" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="242" name="bottleneck2_7/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck2_7/add" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="243" name="bottleneck2_8/dim_red/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f32" offset="93340" shape="16, 64, 1, 1" size="4096"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>16</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="244" name="bottleneck2_8/dim_red/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="0, 0" pads_end="0, 0" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>16</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP32">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="245" name="data_add_2395323958" type="Const" version="opset1">
+ <data element_type="f32" offset="97436" shape="1, 16, 1, 1" size="64"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="246" name="bottleneck2_8/dim_red/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck2_8/dim_red/conv" precision="FP32">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="247" name="bottleneck2_8/dim_red/fn/weights3100840364" type="Const" version="opset1">
+ <data element_type="f32" offset="4664" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="248" name="bottleneck2_8/dim_red/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck2_8/dim_red/conv" precision="FP32">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="249" name="bottleneck2_8/inner/dw1/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f32" offset="97500" shape="16, 1, 1, 3, 3" size="576"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="250" name="bottleneck2_8/inner/dw1/conv" type="GroupConvolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="1, 1" pads_end="1, 1" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP32">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="251" name="data_add_2396123966" type="Const" version="opset1">
+ <data element_type="f32" offset="98076" shape="1, 16, 1, 1" size="64"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="252" name="bottleneck2_8/inner/dw1/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck2_8/inner/dw1/conv" precision="FP32">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="253" name="bottleneck2_8/inner/dw1/fn/weights3094839839" type="Const" version="opset1">
+ <data element_type="f32" offset="4664" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="254" name="bottleneck2_8/inner/dw1/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck2_8/inner/dw1/conv" precision="FP32">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="255" name="bottleneck2_8/dim_inc/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f32" offset="98140" shape="64, 16, 1, 1" size="4096"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>64</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="256" name="bottleneck2_8/dim_inc/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="0, 0" pads_end="0, 0" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>16</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>64</dim>
+ <dim>16</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="257" name="data_add_2396923974" type="Const" version="opset1">
+ <data element_type="f32" offset="102236" shape="1, 64, 1, 1" size="256"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="258" name="bottleneck2_8/dim_inc/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck2_8/dim_inc/conv" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="259" name="bottleneck2_8/add" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck2_8/add" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="260" name="bottleneck2_8/fn/weights3106440124" type="Const" version="opset1">
+ <data element_type="f32" offset="4664" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="261" name="bottleneck2_8/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck2_8/add" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="262" name="bottleneck3_0/skip/pooling" type="MaxPool" version="opset1">
+ <data auto_pad="explicit" kernel="2, 2" pads_begin="0, 0" pads_end="0, 0" rounding_type="ceil" strides="2, 2"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </input>
+ <output>
+ <port id="1" names="bottleneck3_0/skip/pooling" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="263" name="bottleneck3_0/skip/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f32" offset="102492" shape="128, 64, 1, 1" size="32768"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>128</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="264" name="bottleneck3_0/skip/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="0, 0" pads_end="0, 0" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>128</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP32">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="265" name="data_add_2397723982" type="Const" version="opset1">
+ <data element_type="f32" offset="135260" shape="1, 128, 1, 1" size="512"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="266" name="bottleneck3_0/skip/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_0/skip/conv" precision="FP32">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="267" name="bottleneck3_0/dim_red/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f32" offset="135772" shape="32, 64, 1, 1" size="8192"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>32</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="268" name="bottleneck3_0/dim_red/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="0, 0" pads_end="0, 0" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>32</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="269" name="data_add_2398523990" type="Const" version="opset1">
+ <data element_type="f32" offset="143964" shape="1, 32, 1, 1" size="128"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="270" name="bottleneck3_0/dim_red/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_0/dim_red/conv" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="271" name="bottleneck3_0/dim_red/fn/weights3097640670" type="Const" version="opset1">
+ <data element_type="f32" offset="4664" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="272" name="bottleneck3_0/dim_red/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_0/dim_red/conv" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="273" name="bottleneck3_0/inner/dw1/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f32" offset="144092" shape="32, 1, 1, 3, 3" size="1152"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="274" name="bottleneck3_0/inner/dw1/conv" type="GroupConvolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="1, 1" pads_end="1, 1" strides="2, 2"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>80</dim>
+ <dim>136</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="275" name="data_add_2399323998" type="Const" version="opset1">
+ <data element_type="f32" offset="145244" shape="1, 32, 1, 1" size="128"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="276" name="bottleneck3_0/inner/dw1/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_0/inner/dw1/conv" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="277" name="bottleneck3_0/inner/dw1/fn/weights3079640607" type="Const" version="opset1">
+ <data element_type="f32" offset="4664" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="278" name="bottleneck3_0/inner/dw1/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_0/inner/dw1/conv" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="279" name="bottleneck3_0/dim_inc/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f32" offset="145372" shape="128, 32, 1, 1" size="16384"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>128</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="280" name="bottleneck3_0/dim_inc/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="0, 0" pads_end="0, 0" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>128</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP32">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="281" name="data_add_2400124006" type="Const" version="opset1">
+ <data element_type="f32" offset="161756" shape="1, 128, 1, 1" size="512"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="282" name="bottleneck3_0/dim_inc/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_0/dim_inc/conv" precision="FP32">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="283" name="bottleneck3_0/add" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_0/add" precision="FP32">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="284" name="bottleneck3_0/fn/weights3080840268" type="Const" version="opset1">
+ <data element_type="f32" offset="4664" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="285" name="bottleneck3_0/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_0/add" precision="FP32">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="286" name="bottleneck3_1/dim_red/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f32" offset="162268" shape="32, 128, 1, 1" size="16384"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>32</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="287" name="bottleneck3_1/dim_red/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="0, 0" pads_end="0, 0" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>32</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="288" name="data_add_2400924014" type="Const" version="opset1">
+ <data element_type="f32" offset="178652" shape="1, 32, 1, 1" size="128"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="289" name="bottleneck3_1/dim_red/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_1/dim_red/conv" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="290" name="bottleneck3_1/dim_red/fn/weights3102040538" type="Const" version="opset1">
+ <data element_type="f32" offset="4664" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="291" name="bottleneck3_1/dim_red/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_1/dim_red/conv" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="292" name="bottleneck3_1/inner/dw1/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f32" offset="178780" shape="32, 1, 1, 3, 3" size="1152"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="293" name="bottleneck3_1/inner/dw1/conv" type="GroupConvolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="1, 1" pads_end="1, 1" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="294" name="data_add_2401724022" type="Const" version="opset1">
+ <data element_type="f32" offset="179932" shape="1, 32, 1, 1" size="128"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="295" name="bottleneck3_1/inner/dw1/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_1/inner/dw1/conv" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="296" name="bottleneck3_1/inner/dw1/fn/weights3082440517" type="Const" version="opset1">
+ <data element_type="f32" offset="4664" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="297" name="bottleneck3_1/inner/dw1/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_1/inner/dw1/conv" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="298" name="bottleneck3_1/dim_inc/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f32" offset="180060" shape="128, 32, 1, 1" size="16384"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>128</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="299" name="bottleneck3_1/dim_inc/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="0, 0" pads_end="0, 0" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>128</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP32">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="300" name="data_add_2402524030" type="Const" version="opset1">
+ <data element_type="f32" offset="196444" shape="1, 128, 1, 1" size="512"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="301" name="bottleneck3_1/dim_inc/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_1/dim_inc/conv" precision="FP32">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="302" name="bottleneck3_1/add" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_1/add" precision="FP32">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="303" name="bottleneck3_1/fn/weights3086039869" type="Const" version="opset1">
+ <data element_type="f32" offset="4664" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="304" name="bottleneck3_1/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_1/add" precision="FP32">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="305" name="bottleneck3_2/dim_red/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f32" offset="196956" shape="32, 128, 1, 1" size="16384"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>32</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="306" name="bottleneck3_2/dim_red/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="0, 0" pads_end="0, 0" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>32</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="307" name="data_add_2403324038" type="Const" version="opset1">
+ <data element_type="f32" offset="213340" shape="1, 32, 1, 1" size="128"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="308" name="bottleneck3_2/dim_red/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_2/dim_red/conv" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="309" name="bottleneck3_2/dim_red/fn/weights3117639980" type="Const" version="opset1">
+ <data element_type="f32" offset="4664" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="310" name="bottleneck3_2/dim_red/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_2/dim_red/conv" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="311" name="bottleneck3_2/inner/dw1/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f32" offset="213468" shape="32, 1, 1, 3, 3" size="1152"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="312" name="bottleneck3_2/inner/dw1/conv" type="GroupConvolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="1, 1" pads_end="1, 1" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="313" name="data_add_2404124046" type="Const" version="opset1">
+ <data element_type="f32" offset="214620" shape="1, 32, 1, 1" size="128"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="314" name="bottleneck3_2/inner/dw1/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_2/inner/dw1/conv" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="315" name="bottleneck3_2/inner/dw1/fn/weights3108039773" type="Const" version="opset1">
+ <data element_type="f32" offset="4664" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="316" name="bottleneck3_2/inner/dw1/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_2/inner/dw1/conv" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="317" name="bottleneck3_2/dim_inc/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f32" offset="214748" shape="128, 32, 1, 1" size="16384"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>128</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="318" name="bottleneck3_2/dim_inc/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="0, 0" pads_end="0, 0" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>128</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP32">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="319" name="data_add_2404924054" type="Const" version="opset1">
+ <data element_type="f32" offset="231132" shape="1, 128, 1, 1" size="512"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="320" name="bottleneck3_2/dim_inc/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_2/dim_inc/conv" precision="FP32">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="321" name="bottleneck3_2/add" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_2/add" precision="FP32">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="322" name="bottleneck3_2/fn/weights3093640130" type="Const" version="opset1">
+ <data element_type="f32" offset="4664" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="323" name="bottleneck3_2/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_2/add" precision="FP32">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="324" name="bottleneck3_3/dim_red/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f32" offset="231644" shape="32, 128, 1, 1" size="16384"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>32</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="325" name="bottleneck3_3/dim_red/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="0, 0" pads_end="0, 0" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>32</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="326" name="data_add_2405724062" type="Const" version="opset1">
+ <data element_type="f32" offset="248028" shape="1, 32, 1, 1" size="128"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="327" name="bottleneck3_3/dim_red/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_3/dim_red/conv" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="328" name="bottleneck3_3/dim_red/fn/weights3107239758" type="Const" version="opset1">
+ <data element_type="f32" offset="4664" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="329" name="bottleneck3_3/dim_red/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_3/dim_red/conv" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="330" name="bottleneck3_3/inner/dw1/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f32" offset="248156" shape="32, 1, 1, 3, 3" size="1152"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="331" name="bottleneck3_3/inner/dw1/conv" type="GroupConvolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="1, 1" pads_end="1, 1" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="332" name="data_add_2406524070" type="Const" version="opset1">
+ <data element_type="f32" offset="249308" shape="1, 32, 1, 1" size="128"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="333" name="bottleneck3_3/inner/dw1/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_3/inner/dw1/conv" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="334" name="bottleneck3_3/inner/dw1/fn/weights3104440001" type="Const" version="opset1">
+ <data element_type="f32" offset="4664" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="335" name="bottleneck3_3/inner/dw1/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_3/inner/dw1/conv" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="336" name="bottleneck3_3/dim_inc/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f32" offset="249436" shape="128, 32, 1, 1" size="16384"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>128</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="337" name="bottleneck3_3/dim_inc/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="0, 0" pads_end="0, 0" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>128</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP32">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="338" name="data_add_2407324078" type="Const" version="opset1">
+ <data element_type="f32" offset="265820" shape="1, 128, 1, 1" size="512"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="339" name="bottleneck3_3/dim_inc/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_3/dim_inc/conv" precision="FP32">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="340" name="bottleneck3_3/add" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_3/add" precision="FP32">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="341" name="bottleneck3_3/fn/weights3083640325" type="Const" version="opset1">
+ <data element_type="f32" offset="4664" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="342" name="bottleneck3_3/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_3/add" precision="FP32">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="343" name="bottleneck3_4/dim_red/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f32" offset="266332" shape="32, 128, 1, 1" size="16384"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>32</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="344" name="bottleneck3_4/dim_red/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="0, 0" pads_end="0, 0" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>32</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="345" name="data_add_2408124086" type="Const" version="opset1">
+ <data element_type="f32" offset="282716" shape="1, 32, 1, 1" size="128"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="346" name="bottleneck3_4/dim_red/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_4/dim_red/conv" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="347" name="bottleneck3_4/dim_red/fn/weights3077240091" type="Const" version="opset1">
+ <data element_type="f32" offset="4664" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="348" name="bottleneck3_4/dim_red/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_4/dim_red/conv" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="349" name="bottleneck3_4/inner/dw1/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f32" offset="282844" shape="32, 1, 1, 3, 3" size="1152"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="350" name="bottleneck3_4/inner/dw1/conv" type="GroupConvolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="1, 1" pads_end="1, 1" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="351" name="data_add_2408924094" type="Const" version="opset1">
+ <data element_type="f32" offset="283996" shape="1, 32, 1, 1" size="128"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="352" name="bottleneck3_4/inner/dw1/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_4/inner/dw1/conv" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="353" name="bottleneck3_4/inner/dw1/fn/weights3099640157" type="Const" version="opset1">
+ <data element_type="f32" offset="4664" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="354" name="bottleneck3_4/inner/dw1/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_4/inner/dw1/conv" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="355" name="bottleneck3_4/dim_inc/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f32" offset="284124" shape="128, 32, 1, 1" size="16384"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>128</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="356" name="bottleneck3_4/dim_inc/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="0, 0" pads_end="0, 0" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>128</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP32">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="357" name="data_add_2409724102" type="Const" version="opset1">
+ <data element_type="f32" offset="300508" shape="1, 128, 1, 1" size="512"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="358" name="bottleneck3_4/dim_inc/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_4/dim_inc/conv" precision="FP32">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="359" name="bottleneck3_4/add" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_4/add" precision="FP32">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="360" name="bottleneck3_4/fn/weights3105640382" type="Const" version="opset1">
+ <data element_type="f32" offset="4664" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="361" name="bottleneck3_4/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_4/add" precision="FP32">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="362" name="bottleneck3_5/dim_red/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f32" offset="301020" shape="32, 128, 1, 1" size="16384"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>32</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="363" name="bottleneck3_5/dim_red/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="0, 0" pads_end="0, 0" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>32</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="364" name="data_add_2410524110" type="Const" version="opset1">
+ <data element_type="f32" offset="317404" shape="1, 32, 1, 1" size="128"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="365" name="bottleneck3_5/dim_red/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_5/dim_red/conv" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="366" name="bottleneck3_5/dim_red/fn/weights3081240661" type="Const" version="opset1">
+ <data element_type="f32" offset="4664" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="367" name="bottleneck3_5/dim_red/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_5/dim_red/conv" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="368" name="bottleneck3_5/inner/dw1/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f32" offset="317532" shape="32, 1, 1, 3, 3" size="1152"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="369" name="bottleneck3_5/inner/dw1/conv" type="GroupConvolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="1, 1" pads_end="1, 1" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="370" name="data_add_2411324118" type="Const" version="opset1">
+ <data element_type="f32" offset="318684" shape="1, 32, 1, 1" size="128"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="371" name="bottleneck3_5/inner/dw1/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_5/inner/dw1/conv" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="372" name="bottleneck3_5/inner/dw1/fn/weights3113240100" type="Const" version="opset1">
+ <data element_type="f32" offset="4664" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="373" name="bottleneck3_5/inner/dw1/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_5/inner/dw1/conv" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="374" name="bottleneck3_5/dim_inc/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f32" offset="318812" shape="128, 32, 1, 1" size="16384"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>128</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="375" name="bottleneck3_5/dim_inc/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="0, 0" pads_end="0, 0" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>128</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP32">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="376" name="data_add_2412124126" type="Const" version="opset1">
+ <data element_type="f32" offset="335196" shape="1, 128, 1, 1" size="512"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="377" name="bottleneck3_5/dim_inc/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_5/dim_inc/conv" precision="FP32">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="378" name="bottleneck3_5/add" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_5/add" precision="FP32">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="379" name="bottleneck3_5/fn/weights3108439911" type="Const" version="opset1">
+ <data element_type="f32" offset="4664" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="380" name="bottleneck3_5/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_5/add" precision="FP32">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="381" name="bottleneck3_6/dim_red/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f32" offset="335708" shape="32, 128, 1, 1" size="16384"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>32</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="382" name="bottleneck3_6/dim_red/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="0, 0" pads_end="0, 0" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>32</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="383" name="data_add_2412924134" type="Const" version="opset1">
+ <data element_type="f32" offset="352092" shape="1, 32, 1, 1" size="128"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="384" name="bottleneck3_6/dim_red/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_6/dim_red/conv" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="385" name="bottleneck3_6/dim_red/fn/weights3084040604" type="Const" version="opset1">
+ <data element_type="f32" offset="4664" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="386" name="bottleneck3_6/dim_red/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_6/dim_red/conv" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="387" name="bottleneck3_6/inner/dw1/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f32" offset="352220" shape="32, 1, 1, 3, 3" size="1152"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="388" name="bottleneck3_6/inner/dw1/conv" type="GroupConvolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="1, 1" pads_end="1, 1" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="389" name="data_add_2413724142" type="Const" version="opset1">
+ <data element_type="f32" offset="353372" shape="1, 32, 1, 1" size="128"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="390" name="bottleneck3_6/inner/dw1/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_6/inner/dw1/conv" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="391" name="bottleneck3_6/inner/dw1/fn/weights3090840310" type="Const" version="opset1">
+ <data element_type="f32" offset="4664" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="392" name="bottleneck3_6/inner/dw1/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_6/inner/dw1/conv" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="393" name="bottleneck3_6/dim_inc/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f32" offset="353500" shape="128, 32, 1, 1" size="16384"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>128</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="394" name="bottleneck3_6/dim_inc/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="0, 0" pads_end="0, 0" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>128</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP32">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="395" name="data_add_2414524150" type="Const" version="opset1">
+ <data element_type="f32" offset="369884" shape="1, 128, 1, 1" size="512"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="396" name="bottleneck3_6/dim_inc/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_6/dim_inc/conv" precision="FP32">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="397" name="bottleneck3_6/add" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_6/add" precision="FP32">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="398" name="bottleneck3_6/fn/weights3090039914" type="Const" version="opset1">
+ <data element_type="f32" offset="4664" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="399" name="bottleneck3_6/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_6/add" precision="FP32">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="400" name="bottleneck3_7/dim_red/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f32" offset="370396" shape="32, 128, 1, 1" size="16384"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>32</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="401" name="bottleneck3_7/dim_red/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="0, 0" pads_end="0, 0" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>32</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="402" name="data_add_2415324158" type="Const" version="opset1">
+ <data element_type="f32" offset="386780" shape="1, 32, 1, 1" size="128"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="403" name="bottleneck3_7/dim_red/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_7/dim_red/conv" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="404" name="bottleneck3_7/dim_red/fn/weights3113640679" type="Const" version="opset1">
+ <data element_type="f32" offset="4664" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="405" name="bottleneck3_7/dim_red/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_7/dim_red/conv" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="406" name="bottleneck3_7/inner/dw1/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f32" offset="386908" shape="32, 1, 1, 3, 3" size="1152"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="407" name="bottleneck3_7/inner/dw1/conv" type="GroupConvolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="1, 1" pads_end="1, 1" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="408" name="data_add_2416124166" type="Const" version="opset1">
+ <data element_type="f32" offset="388060" shape="1, 32, 1, 1" size="128"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="409" name="bottleneck3_7/inner/dw1/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_7/inner/dw1/conv" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="410" name="bottleneck3_7/inner/dw1/fn/weights3098040349" type="Const" version="opset1">
+ <data element_type="f32" offset="4664" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="411" name="bottleneck3_7/inner/dw1/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_7/inner/dw1/conv" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="412" name="bottleneck3_7/dim_inc/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f32" offset="388188" shape="128, 32, 1, 1" size="16384"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>128</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="413" name="bottleneck3_7/dim_inc/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="0, 0" pads_end="0, 0" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>128</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP32">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="414" name="data_add_2416924174" type="Const" version="opset1">
+ <data element_type="f32" offset="404572" shape="1, 128, 1, 1" size="512"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="415" name="bottleneck3_7/dim_inc/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_7/dim_inc/conv" precision="FP32">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="416" name="bottleneck3_7/add" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_7/add" precision="FP32">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="417" name="bottleneck3_7/fn/weights3102840676" type="Const" version="opset1">
+ <data element_type="f32" offset="4664" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="418" name="bottleneck3_7/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_7/add" precision="FP32">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="419" name="bottleneck3_8/dim_red/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f32" offset="405084" shape="32, 128, 1, 1" size="16384"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>32</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="420" name="bottleneck3_8/dim_red/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="0, 0" pads_end="0, 0" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>32</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="421" name="data_add_2417724182" type="Const" version="opset1">
+ <data element_type="f32" offset="421468" shape="1, 32, 1, 1" size="128"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="422" name="bottleneck3_8/dim_red/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_8/dim_red/conv" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="423" name="bottleneck3_8/dim_red/fn/weights3098440022" type="Const" version="opset1">
+ <data element_type="f32" offset="4664" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="424" name="bottleneck3_8/dim_red/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_8/dim_red/conv" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="425" name="bottleneck3_8/inner/dw1/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f32" offset="421596" shape="32, 1, 1, 3, 3" size="1152"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="426" name="bottleneck3_8/inner/dw1/conv" type="GroupConvolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="1, 1" pads_end="1, 1" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="427" name="data_add_2418524190" type="Const" version="opset1">
+ <data element_type="f32" offset="422748" shape="1, 32, 1, 1" size="128"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="428" name="bottleneck3_8/inner/dw1/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_8/inner/dw1/conv" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="429" name="bottleneck3_8/inner/dw1/fn/weights3104839899" type="Const" version="opset1">
+ <data element_type="f32" offset="4664" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="430" name="bottleneck3_8/inner/dw1/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_8/inner/dw1/conv" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="431" name="bottleneck3_8/dim_inc/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f32" offset="422876" shape="128, 32, 1, 1" size="16384"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>128</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="432" name="bottleneck3_8/dim_inc/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="0, 0" pads_end="0, 0" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>128</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP32">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="433" name="data_add_2419324198" type="Const" version="opset1">
+ <data element_type="f32" offset="439260" shape="1, 128, 1, 1" size="512"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="434" name="bottleneck3_8/dim_inc/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_8/dim_inc/conv" precision="FP32">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="435" name="bottleneck3_8/add" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_8/add" precision="FP32">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="436" name="bottleneck3_8/fn/weights3083239761" type="Const" version="opset1">
+ <data element_type="f32" offset="4664" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="437" name="bottleneck3_8/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_8/add" precision="FP32">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="438" name="bottleneck3_9/dim_red/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f32" offset="439772" shape="32, 128, 1, 1" size="16384"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>32</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="439" name="bottleneck3_9/dim_red/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="0, 0" pads_end="0, 0" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>32</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="440" name="data_add_2420124206" type="Const" version="opset1">
+ <data element_type="f32" offset="456156" shape="1, 32, 1, 1" size="128"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="441" name="bottleneck3_9/dim_red/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_9/dim_red/conv" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="442" name="bottleneck3_9/dim_red/fn/weights3093240487" type="Const" version="opset1">
+ <data element_type="f32" offset="4664" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="443" name="bottleneck3_9/dim_red/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_9/dim_red/conv" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="444" name="bottleneck3_9/inner/dw1/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f32" offset="456284" shape="32, 1, 1, 3, 3" size="1152"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="445" name="bottleneck3_9/inner/dw1/conv" type="GroupConvolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="1, 1" pads_end="1, 1" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="446" name="data_add_2420924214" type="Const" version="opset1">
+ <data element_type="f32" offset="457436" shape="1, 32, 1, 1" size="128"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="447" name="bottleneck3_9/inner/dw1/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_9/inner/dw1/conv" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="448" name="bottleneck3_9/inner/dw1/fn/weights3112040592" type="Const" version="opset1">
+ <data element_type="f32" offset="4664" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="449" name="bottleneck3_9/inner/dw1/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_9/inner/dw1/conv" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="450" name="bottleneck3_9/dim_inc/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f32" offset="457564" shape="128, 32, 1, 1" size="16384"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>128</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="451" name="bottleneck3_9/dim_inc/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="0, 0" pads_end="0, 0" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>128</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP32">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="452" name="data_add_2421724222" type="Const" version="opset1">
+ <data element_type="f32" offset="473948" shape="1, 128, 1, 1" size="512"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="453" name="bottleneck3_9/dim_inc/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_9/dim_inc/conv" precision="FP32">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="454" name="bottleneck3_9/add" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_9/add" precision="FP32">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="455" name="bottleneck3_9/fn/weights3116840514" type="Const" version="opset1">
+ <data element_type="f32" offset="4664" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="456" name="bottleneck3_9/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_9/add" precision="FP32">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="457" name="bottleneck3_10/dim_red/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f32" offset="474460" shape="32, 128, 1, 1" size="16384"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>32</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="458" name="bottleneck3_10/dim_red/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="0, 0" pads_end="0, 0" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>32</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="459" name="data_add_2422524230" type="Const" version="opset1">
+ <data element_type="f32" offset="490844" shape="1, 32, 1, 1" size="128"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="460" name="bottleneck3_10/dim_red/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_10/dim_red/conv" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="461" name="bottleneck3_10/dim_red/fn/weights3100040610" type="Const" version="opset1">
+ <data element_type="f32" offset="4664" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="462" name="bottleneck3_10/dim_red/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_10/dim_red/conv" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="463" name="bottleneck3_10/inner/dw1/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f32" offset="490972" shape="32, 1, 1, 3, 3" size="1152"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="464" name="bottleneck3_10/inner/dw1/conv" type="GroupConvolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="1, 1" pads_end="1, 1" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="465" name="data_add_2423324238" type="Const" version="opset1">
+ <data element_type="f32" offset="492124" shape="1, 32, 1, 1" size="128"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="466" name="bottleneck3_10/inner/dw1/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_10/inner/dw1/conv" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="467" name="bottleneck3_10/inner/dw1/fn/weights3116040688" type="Const" version="opset1">
+ <data element_type="f32" offset="4664" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="468" name="bottleneck3_10/inner/dw1/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_10/inner/dw1/conv" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="469" name="bottleneck3_10/dim_inc/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f32" offset="492252" shape="128, 32, 1, 1" size="16384"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>128</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="470" name="bottleneck3_10/dim_inc/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="0, 0" pads_end="0, 0" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>32</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>128</dim>
+ <dim>32</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP32">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="471" name="data_add_2424124246" type="Const" version="opset1">
+ <data element_type="f32" offset="508636" shape="1, 128, 1, 1" size="512"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="472" name="bottleneck3_10/dim_inc/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_10/dim_inc/conv" precision="FP32">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="473" name="bottleneck3_10/add" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_10/add" precision="FP32">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="474" name="bottleneck3_10/fn/weights3103639695" type="Const" version="opset1">
+ <data element_type="f32" offset="4664" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="475" name="bottleneck3_10/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck3_10/add" precision="FP32">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="476" name="bottleneck4_0/skip/pooling" type="MaxPool" version="opset1">
+ <data auto_pad="explicit" kernel="2, 2" pads_begin="0, 0" pads_end="0, 0" rounding_type="ceil" strides="2, 2"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </input>
+ <output>
+ <port id="1" names="bottleneck4_0/skip/pooling" precision="FP32">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="477" name="bottleneck4_0/skip/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f32" offset="509148" shape="256, 128, 1, 1" size="131072"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>256</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="478" name="bottleneck4_0/skip/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="0, 0" pads_end="0, 0" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>256</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP32">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="479" name="data_add_2424924254" type="Const" version="opset1">
+ <data element_type="f32" offset="640220" shape="1, 256, 1, 1" size="1024"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="480" name="bottleneck4_0/skip/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_0/skip/conv" precision="FP32">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="481" name="bottleneck4_0/dim_red/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f32" offset="641244" shape="64, 128, 1, 1" size="32768"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>64</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="482" name="bottleneck4_0/dim_red/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="0, 0" pads_end="0, 0" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>128</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>64</dim>
+ <dim>128</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="483" name="data_add_2425724262" type="Const" version="opset1">
+ <data element_type="f32" offset="674012" shape="1, 64, 1, 1" size="256"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="484" name="bottleneck4_0/dim_red/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_0/dim_red/conv" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="485" name="bottleneck4_0/dim_red/fn/weights3109639941" type="Const" version="opset1">
+ <data element_type="f32" offset="4664" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="486" name="bottleneck4_0/dim_red/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_0/dim_red/conv" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="487" name="bottleneck4_0/inner/dw1/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f32" offset="674268" shape="64, 1, 1, 3, 3" size="2304"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="488" name="bottleneck4_0/inner/dw1/conv" type="GroupConvolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="1, 1" pads_end="1, 1" strides="2, 2"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>40</dim>
+ <dim>68</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="489" name="data_add_2426524270" type="Const" version="opset1">
+ <data element_type="f32" offset="676572" shape="1, 64, 1, 1" size="256"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="490" name="bottleneck4_0/inner/dw1/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_0/inner/dw1/conv" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="491" name="bottleneck4_0/inner/dw1/fn/weights3118439713" type="Const" version="opset1">
+ <data element_type="f32" offset="4664" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="492" name="bottleneck4_0/inner/dw1/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_0/inner/dw1/conv" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="493" name="bottleneck4_0/dim_inc/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f32" offset="676828" shape="256, 64, 1, 1" size="65536"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>256</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="494" name="bottleneck4_0/dim_inc/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="0, 0" pads_end="0, 0" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>256</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP32">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="495" name="data_add_2427324278" type="Const" version="opset1">
+ <data element_type="f32" offset="742364" shape="1, 256, 1, 1" size="1024"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="496" name="bottleneck4_0/dim_inc/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_0/dim_inc/conv" precision="FP32">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="497" name="bottleneck4_0/add" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_0/add" precision="FP32">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="498" name="bottleneck4_0/fn/weights3078039842" type="Const" version="opset1">
+ <data element_type="f32" offset="4664" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="499" name="bottleneck4_0/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_0/add" precision="FP32">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="500" name="bottleneck4_1/dim_red/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f32" offset="743388" shape="64, 256, 1, 1" size="65536"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>64</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="501" name="bottleneck4_1/dim_red/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="0, 0" pads_end="0, 0" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>64</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="502" name="data_add_2428124286" type="Const" version="opset1">
+ <data element_type="f32" offset="808924" shape="1, 64, 1, 1" size="256"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="503" name="bottleneck4_1/dim_red/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_1/dim_red/conv" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="504" name="bottleneck4_1/dim_red/fn/weights3112840550" type="Const" version="opset1">
+ <data element_type="f32" offset="4664" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="505" name="bottleneck4_1/dim_red/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_1/dim_red/conv" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="506" name="bottleneck4_1/inner/dw1/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f32" offset="809180" shape="64, 1, 1, 3, 3" size="2304"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="507" name="bottleneck4_1/inner/dw1/conv" type="GroupConvolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="1, 1" pads_end="1, 1" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="508" name="data_add_2428924294" type="Const" version="opset1">
+ <data element_type="f32" offset="811484" shape="1, 64, 1, 1" size="256"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="509" name="bottleneck4_1/inner/dw1/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_1/inner/dw1/conv" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="510" name="bottleneck4_1/inner/dw1/fn/weights3101640097" type="Const" version="opset1">
+ <data element_type="f32" offset="4664" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="511" name="bottleneck4_1/inner/dw1/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_1/inner/dw1/conv" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="512" name="bottleneck4_1/dim_inc/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f32" offset="811740" shape="256, 64, 1, 1" size="65536"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>256</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="513" name="bottleneck4_1/dim_inc/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="0, 0" pads_end="0, 0" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>256</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP32">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="514" name="data_add_2429724302" type="Const" version="opset1">
+ <data element_type="f32" offset="877276" shape="1, 256, 1, 1" size="1024"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="515" name="bottleneck4_1/dim_inc/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_1/dim_inc/conv" precision="FP32">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="516" name="bottleneck4_1/add" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_1/add" precision="FP32">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="517" name="bottleneck4_1/fn/weights3078840664" type="Const" version="opset1">
+ <data element_type="f32" offset="4664" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="518" name="bottleneck4_1/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_1/add" precision="FP32">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="519" name="bottleneck4_2/dim_red/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f32" offset="878300" shape="64, 256, 1, 1" size="65536"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>64</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="520" name="bottleneck4_2/dim_red/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="0, 0" pads_end="0, 0" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>64</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="521" name="data_add_2430524310" type="Const" version="opset1">
+ <data element_type="f32" offset="943836" shape="1, 64, 1, 1" size="256"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="522" name="bottleneck4_2/dim_red/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_2/dim_red/conv" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="523" name="bottleneck4_2/dim_red/fn/weights3080440121" type="Const" version="opset1">
+ <data element_type="f32" offset="4664" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="524" name="bottleneck4_2/dim_red/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_2/dim_red/conv" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="525" name="bottleneck4_2/inner/dw1/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f32" offset="944092" shape="64, 1, 1, 3, 3" size="2304"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="526" name="bottleneck4_2/inner/dw1/conv" type="GroupConvolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="1, 1" pads_end="1, 1" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="527" name="data_add_2431324318" type="Const" version="opset1">
+ <data element_type="f32" offset="946396" shape="1, 64, 1, 1" size="256"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="528" name="bottleneck4_2/inner/dw1/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_2/inner/dw1/conv" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="529" name="bottleneck4_2/inner/dw1/fn/weights3078440649" type="Const" version="opset1">
+ <data element_type="f32" offset="4664" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="530" name="bottleneck4_2/inner/dw1/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_2/inner/dw1/conv" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="531" name="bottleneck4_2/dim_inc/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f32" offset="946652" shape="256, 64, 1, 1" size="65536"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>256</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="532" name="bottleneck4_2/dim_inc/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="0, 0" pads_end="0, 0" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>256</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP32">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="533" name="data_add_2432124326" type="Const" version="opset1">
+ <data element_type="f32" offset="1012188" shape="1, 256, 1, 1" size="1024"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="534" name="bottleneck4_2/dim_inc/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_2/dim_inc/conv" precision="FP32">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="535" name="bottleneck4_2/add" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_2/add" precision="FP32">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="536" name="bottleneck4_2/fn/weights3084440229" type="Const" version="opset1">
+ <data element_type="f32" offset="4664" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="537" name="bottleneck4_2/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_2/add" precision="FP32">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="538" name="bottleneck4_3/dim_red/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f32" offset="1013212" shape="64, 256, 1, 1" size="65536"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>64</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="539" name="bottleneck4_3/dim_red/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="0, 0" pads_end="0, 0" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>64</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="540" name="data_add_2432924334" type="Const" version="opset1">
+ <data element_type="f32" offset="1078748" shape="1, 64, 1, 1" size="256"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="541" name="bottleneck4_3/dim_red/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_3/dim_red/conv" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="542" name="bottleneck4_3/dim_red/fn/weights3112440511" type="Const" version="opset1">
+ <data element_type="f32" offset="4664" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="543" name="bottleneck4_3/dim_red/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_3/dim_red/conv" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="544" name="bottleneck4_3/inner/dw1/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f32" offset="1079004" shape="64, 1, 1, 3, 3" size="2304"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="545" name="bottleneck4_3/inner/dw1/conv" type="GroupConvolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="1, 1" pads_end="1, 1" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="546" name="data_add_2433724342" type="Const" version="opset1">
+ <data element_type="f32" offset="1081308" shape="1, 64, 1, 1" size="256"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="547" name="bottleneck4_3/inner/dw1/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_3/inner/dw1/conv" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="548" name="bottleneck4_3/inner/dw1/fn/weights3108840466" type="Const" version="opset1">
+ <data element_type="f32" offset="4664" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="549" name="bottleneck4_3/inner/dw1/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_3/inner/dw1/conv" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="550" name="bottleneck4_3/dim_inc/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f32" offset="1081564" shape="256, 64, 1, 1" size="65536"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>256</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="551" name="bottleneck4_3/dim_inc/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="0, 0" pads_end="0, 0" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>256</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP32">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="552" name="data_add_2434524350" type="Const" version="opset1">
+ <data element_type="f32" offset="1147100" shape="1, 256, 1, 1" size="1024"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="553" name="bottleneck4_3/dim_inc/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_3/dim_inc/conv" precision="FP32">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="554" name="bottleneck4_3/add" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_3/add" precision="FP32">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="555" name="bottleneck4_3/fn/weights3088440685" type="Const" version="opset1">
+ <data element_type="f32" offset="4664" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="556" name="bottleneck4_3/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_3/add" precision="FP32">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="557" name="bottleneck4_4/dim_red/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f32" offset="1148124" shape="64, 256, 1, 1" size="65536"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>64</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="558" name="bottleneck4_4/dim_red/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="0, 0" pads_end="0, 0" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>64</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="559" name="data_add_2435324358" type="Const" version="opset1">
+ <data element_type="f32" offset="1213660" shape="1, 64, 1, 1" size="256"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="560" name="bottleneck4_4/dim_red/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_4/dim_red/conv" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="561" name="bottleneck4_4/dim_red/fn/weights3116439731" type="Const" version="opset1">
+ <data element_type="f32" offset="4664" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="562" name="bottleneck4_4/dim_red/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_4/dim_red/conv" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="563" name="bottleneck4_4/inner/dw1/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f32" offset="1213916" shape="64, 1, 1, 3, 3" size="2304"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="564" name="bottleneck4_4/inner/dw1/conv" type="GroupConvolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="1, 1" pads_end="1, 1" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="565" name="data_add_2436124366" type="Const" version="opset1">
+ <data element_type="f32" offset="1216220" shape="1, 64, 1, 1" size="256"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="566" name="bottleneck4_4/inner/dw1/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_4/inner/dw1/conv" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="567" name="bottleneck4_4/inner/dw1/fn/weights3076040484" type="Const" version="opset1">
+ <data element_type="f32" offset="4664" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="568" name="bottleneck4_4/inner/dw1/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_4/inner/dw1/conv" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="569" name="bottleneck4_4/dim_inc/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f32" offset="1216476" shape="256, 64, 1, 1" size="65536"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>256</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="570" name="bottleneck4_4/dim_inc/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="0, 0" pads_end="0, 0" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>256</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP32">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="571" name="data_add_2436924374" type="Const" version="opset1">
+ <data element_type="f32" offset="1282012" shape="1, 256, 1, 1" size="1024"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="572" name="bottleneck4_4/dim_inc/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_4/dim_inc/conv" precision="FP32">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="573" name="bottleneck4_4/add" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_4/add" precision="FP32">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="574" name="bottleneck4_4/fn/weights3098839926" type="Const" version="opset1">
+ <data element_type="f32" offset="4664" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="575" name="bottleneck4_4/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_4/add" precision="FP32">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="576" name="bottleneck4_5/dim_red/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f32" offset="1283036" shape="64, 256, 1, 1" size="65536"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>64</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="577" name="bottleneck4_5/dim_red/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="0, 0" pads_end="0, 0" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>64</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="578" name="data_add_2437724382" type="Const" version="opset1">
+ <data element_type="f32" offset="1348572" shape="1, 64, 1, 1" size="256"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="579" name="bottleneck4_5/dim_red/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_5/dim_red/conv" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="580" name="bottleneck4_5/dim_red/fn/weights3117240481" type="Const" version="opset1">
+ <data element_type="f32" offset="4664" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="581" name="bottleneck4_5/dim_red/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_5/dim_red/conv" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="582" name="bottleneck4_5/inner/dw1/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f32" offset="1348828" shape="64, 1, 1, 3, 3" size="2304"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="583" name="bottleneck4_5/inner/dw1/conv" type="GroupConvolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="1, 1" pads_end="1, 1" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="584" name="data_add_2438524390" type="Const" version="opset1">
+ <data element_type="f32" offset="1351132" shape="1, 64, 1, 1" size="256"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="585" name="bottleneck4_5/inner/dw1/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_5/inner/dw1/conv" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="586" name="bottleneck4_5/inner/dw1/fn/weights3076439938" type="Const" version="opset1">
+ <data element_type="f32" offset="4664" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="587" name="bottleneck4_5/inner/dw1/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_5/inner/dw1/conv" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="588" name="bottleneck4_5/dim_inc/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f32" offset="1351388" shape="256, 64, 1, 1" size="65536"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>256</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="589" name="bottleneck4_5/dim_inc/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="0, 0" pads_end="0, 0" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>256</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP32">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="590" name="data_add_2439324398" type="Const" version="opset1">
+ <data element_type="f32" offset="1416924" shape="1, 256, 1, 1" size="1024"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="591" name="bottleneck4_5/dim_inc/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_5/dim_inc/conv" precision="FP32">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="592" name="bottleneck4_5/add" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_5/add" precision="FP32">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="593" name="bottleneck4_5/fn/weights3084840586" type="Const" version="opset1">
+ <data element_type="f32" offset="4664" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="594" name="bottleneck4_5/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_5/add" precision="FP32">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="595" name="bottleneck4_6/dim_red/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f32" offset="1417948" shape="64, 256, 1, 1" size="65536"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>64</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="596" name="bottleneck4_6/dim_red/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="0, 0" pads_end="0, 0" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>64</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="597" name="data_add_2440124406" type="Const" version="opset1">
+ <data element_type="f32" offset="1483484" shape="1, 64, 1, 1" size="256"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="598" name="bottleneck4_6/dim_red/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_6/dim_red/conv" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="599" name="bottleneck4_6/dim_red/fn/weights3096839815" type="Const" version="opset1">
+ <data element_type="f32" offset="4664" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="600" name="bottleneck4_6/dim_red/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_6/dim_red/conv" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="601" name="bottleneck4_6/inner/dw1/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f32" offset="1483740" shape="64, 1, 1, 3, 3" size="2304"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="602" name="bottleneck4_6/inner/dw1/conv" type="GroupConvolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="1, 1" pads_end="1, 1" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="603" name="data_add_2440924414" type="Const" version="opset1">
+ <data element_type="f32" offset="1486044" shape="1, 64, 1, 1" size="256"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="604" name="bottleneck4_6/inner/dw1/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_6/inner/dw1/conv" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="605" name="bottleneck4_6/inner/dw1/fn/weights3109240031" type="Const" version="opset1">
+ <data element_type="f32" offset="4664" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="606" name="bottleneck4_6/inner/dw1/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_6/inner/dw1/conv" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="607" name="bottleneck4_6/dim_inc/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f32" offset="1486300" shape="256, 64, 1, 1" size="65536"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>256</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="608" name="bottleneck4_6/dim_inc/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="0, 0" pads_end="0, 0" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>256</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP32">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="609" name="data_add_2441724422" type="Const" version="opset1">
+ <data element_type="f32" offset="1551836" shape="1, 256, 1, 1" size="1024"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="610" name="bottleneck4_6/dim_inc/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_6/dim_inc/conv" precision="FP32">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="611" name="bottleneck4_6/add" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_6/add" precision="FP32">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="612" name="bottleneck4_6/fn/weights3111640460" type="Const" version="opset1">
+ <data element_type="f32" offset="4664" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="613" name="bottleneck4_6/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_6/add" precision="FP32">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="614" name="bottleneck4_7/dim_red/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f32" offset="1552860" shape="64, 256, 1, 1" size="65536"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>64</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="615" name="bottleneck4_7/dim_red/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="0, 0" pads_end="0, 0" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>64</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="616" name="data_add_2442524430" type="Const" version="opset1">
+ <data element_type="f32" offset="1618396" shape="1, 64, 1, 1" size="256"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="617" name="bottleneck4_7/dim_red/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_7/dim_red/conv" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="618" name="bottleneck4_7/dim_red/fn/weights3095240595" type="Const" version="opset1">
+ <data element_type="f32" offset="4664" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="619" name="bottleneck4_7/dim_red/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_7/dim_red/conv" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="620" name="bottleneck4_7/inner/dw1/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f32" offset="1618652" shape="64, 1, 1, 3, 3" size="2304"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="621" name="bottleneck4_7/inner/dw1/conv" type="GroupConvolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="1, 1" pads_end="1, 1" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="622" name="data_add_2443324438" type="Const" version="opset1">
+ <data element_type="f32" offset="1620956" shape="1, 64, 1, 1" size="256"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="623" name="bottleneck4_7/inner/dw1/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_7/inner/dw1/conv" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="624" name="bottleneck4_7/inner/dw1/fn/weights3101240547" type="Const" version="opset1">
+ <data element_type="f32" offset="4664" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="625" name="bottleneck4_7/inner/dw1/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_7/inner/dw1/conv" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="626" name="bottleneck4_7/dim_inc/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f32" offset="1621212" shape="256, 64, 1, 1" size="65536"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>256</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="627" name="bottleneck4_7/dim_inc/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="0, 0" pads_end="0, 0" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>256</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP32">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="628" name="data_add_2444124446" type="Const" version="opset1">
+ <data element_type="f32" offset="1686748" shape="1, 256, 1, 1" size="1024"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="629" name="bottleneck4_7/dim_inc/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_7/dim_inc/conv" precision="FP32">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="630" name="bottleneck4_7/add" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_7/add" precision="FP32">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="631" name="bottleneck4_7/fn/weights3118840052" type="Const" version="opset1">
+ <data element_type="f32" offset="4664" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="632" name="bottleneck4_7/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_7/add" precision="FP32">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="633" name="bottleneck4_8/dim_red/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f32" offset="1687772" shape="64, 256, 1, 1" size="65536"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>64</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="634" name="bottleneck4_8/dim_red/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="0, 0" pads_end="0, 0" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>64</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="635" name="data_add_2444924454" type="Const" version="opset1">
+ <data element_type="f32" offset="1753308" shape="1, 64, 1, 1" size="256"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="636" name="bottleneck4_8/dim_red/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_8/dim_red/conv" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="637" name="bottleneck4_8/dim_red/fn/weights3091640496" type="Const" version="opset1">
+ <data element_type="f32" offset="4664" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="638" name="bottleneck4_8/dim_red/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_8/dim_red/conv" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="639" name="bottleneck4_8/inner/dw1/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f32" offset="1753564" shape="64, 1, 1, 3, 3" size="2304"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="640" name="bottleneck4_8/inner/dw1/conv" type="GroupConvolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="1, 1" pads_end="1, 1" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="641" name="data_add_2445724462" type="Const" version="opset1">
+ <data element_type="f32" offset="1755868" shape="1, 64, 1, 1" size="256"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="642" name="bottleneck4_8/inner/dw1/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_8/inner/dw1/conv" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="643" name="bottleneck4_8/inner/dw1/fn/weights3094039866" type="Const" version="opset1">
+ <data element_type="f32" offset="4664" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="644" name="bottleneck4_8/inner/dw1/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_8/inner/dw1/conv" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="645" name="bottleneck4_8/dim_inc/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f32" offset="1756124" shape="256, 64, 1, 1" size="65536"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>256</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="646" name="bottleneck4_8/dim_inc/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="0, 0" pads_end="0, 0" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>256</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP32">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="647" name="data_add_2446524470" type="Const" version="opset1">
+ <data element_type="f32" offset="1821660" shape="1, 256, 1, 1" size="1024"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="648" name="bottleneck4_8/dim_inc/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_8/dim_inc/conv" precision="FP32">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="649" name="bottleneck4_8/add" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_8/add" precision="FP32">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="650" name="bottleneck4_8/fn/weights3089640028" type="Const" version="opset1">
+ <data element_type="f32" offset="4664" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="651" name="bottleneck4_8/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_8/add" precision="FP32">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="652" name="bottleneck4_9/dim_red/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f32" offset="1822684" shape="64, 256, 1, 1" size="65536"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>64</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="653" name="bottleneck4_9/dim_red/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="0, 0" pads_end="0, 0" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>64</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="654" name="data_add_2447324478" type="Const" version="opset1">
+ <data element_type="f32" offset="1888220" shape="1, 64, 1, 1" size="256"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="655" name="bottleneck4_9/dim_red/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_9/dim_red/conv" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="656" name="bottleneck4_9/dim_red/fn/weights3105240346" type="Const" version="opset1">
+ <data element_type="f32" offset="4664" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="657" name="bottleneck4_9/dim_red/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_9/dim_red/conv" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="658" name="bottleneck4_9/inner/dw1/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f32" offset="1888476" shape="64, 1, 1, 3, 3" size="2304"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="659" name="bottleneck4_9/inner/dw1/conv" type="GroupConvolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="1, 1" pads_end="1, 1" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="660" name="data_add_2448124486" type="Const" version="opset1">
+ <data element_type="f32" offset="1890780" shape="1, 64, 1, 1" size="256"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="661" name="bottleneck4_9/inner/dw1/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_9/inner/dw1/conv" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="662" name="bottleneck4_9/inner/dw1/fn/weights3099240472" type="Const" version="opset1">
+ <data element_type="f32" offset="4664" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="663" name="bottleneck4_9/inner/dw1/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_9/inner/dw1/conv" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="664" name="bottleneck4_9/dim_inc/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f32" offset="1891036" shape="256, 64, 1, 1" size="65536"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>256</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="665" name="bottleneck4_9/dim_inc/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="0, 0" pads_end="0, 0" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>256</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP32">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="666" name="data_add_2448924494" type="Const" version="opset1">
+ <data element_type="f32" offset="1956572" shape="1, 256, 1, 1" size="1024"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="667" name="bottleneck4_9/dim_inc/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_9/dim_inc/conv" precision="FP32">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="668" name="bottleneck4_9/add" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_9/add" precision="FP32">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="669" name="bottleneck4_9/fn/weights3092040463" type="Const" version="opset1">
+ <data element_type="f32" offset="4664" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="670" name="bottleneck4_9/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_9/add" precision="FP32">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="671" name="bottleneck4_10/dim_red/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f32" offset="1957596" shape="64, 256, 1, 1" size="65536"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>64</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="672" name="bottleneck4_10/dim_red/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="0, 0" pads_end="0, 0" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>64</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="673" name="data_add_2449724502" type="Const" version="opset1">
+ <data element_type="f32" offset="2023132" shape="1, 64, 1, 1" size="256"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="674" name="bottleneck4_10/dim_red/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_10/dim_red/conv" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="675" name="bottleneck4_10/dim_red/fn/weights3086840214" type="Const" version="opset1">
+ <data element_type="f32" offset="4664" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="676" name="bottleneck4_10/dim_red/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_10/dim_red/conv" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="677" name="bottleneck4_10/inner/dw1/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f32" offset="2023388" shape="64, 1, 1, 3, 3" size="2304"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="678" name="bottleneck4_10/inner/dw1/conv" type="GroupConvolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="1, 1" pads_end="1, 1" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="679" name="data_add_2450524510" type="Const" version="opset1">
+ <data element_type="f32" offset="2025692" shape="1, 64, 1, 1" size="256"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="680" name="bottleneck4_10/inner/dw1/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_10/inner/dw1/conv" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="681" name="bottleneck4_10/inner/dw1/fn/weights3111240634" type="Const" version="opset1">
+ <data element_type="f32" offset="4664" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="682" name="bottleneck4_10/inner/dw1/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_10/inner/dw1/conv" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="683" name="bottleneck4_10/dim_inc/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f32" offset="2025948" shape="256, 64, 1, 1" size="65536"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>256</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="684" name="bottleneck4_10/dim_inc/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="0, 0" pads_end="0, 0" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>256</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP32">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="685" name="data_add_2451324518" type="Const" version="opset1">
+ <data element_type="f32" offset="2091484" shape="1, 256, 1, 1" size="1024"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="686" name="bottleneck4_10/dim_inc/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_10/dim_inc/conv" precision="FP32">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="687" name="bottleneck4_10/add" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_10/add" precision="FP32">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="688" name="bottleneck4_10/fn/weights3089240424" type="Const" version="opset1">
+ <data element_type="f32" offset="4664" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="689" name="bottleneck4_10/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_10/add" precision="FP32">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="690" name="bottleneck4_11/dim_red/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f32" offset="2092508" shape="64, 256, 1, 1" size="65536"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>64</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="691" name="bottleneck4_11/dim_red/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="0, 0" pads_end="0, 0" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>64</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="692" name="data_add_2452124526" type="Const" version="opset1">
+ <data element_type="f32" offset="2158044" shape="1, 64, 1, 1" size="256"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="693" name="bottleneck4_11/dim_red/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_11/dim_red/conv" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="694" name="bottleneck4_11/dim_red/fn/weights3104040334" type="Const" version="opset1">
+ <data element_type="f32" offset="4664" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="695" name="bottleneck4_11/dim_red/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_11/dim_red/conv" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="696" name="bottleneck4_11/inner/dw1/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f32" offset="2158300" shape="64, 1, 1, 3, 3" size="2304"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="697" name="bottleneck4_11/inner/dw1/conv" type="GroupConvolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="1, 1" pads_end="1, 1" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="698" name="data_add_2452924534" type="Const" version="opset1">
+ <data element_type="f32" offset="2160604" shape="1, 64, 1, 1" size="256"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="699" name="bottleneck4_11/inner/dw1/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_11/inner/dw1/conv" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="700" name="bottleneck4_11/inner/dw1/fn/weights3114040292" type="Const" version="opset1">
+ <data element_type="f32" offset="4664" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="701" name="bottleneck4_11/inner/dw1/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_11/inner/dw1/conv" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="702" name="bottleneck4_11/dim_inc/bn/mean/Fused_Mul__copy" type="Const" version="opset1">
+ <data element_type="f32" offset="2160860" shape="256, 64, 1, 1" size="65536"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>256</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="703" name="bottleneck4_11/dim_inc/conv" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="0, 0" pads_end="0, 0" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>64</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>256</dim>
+ <dim>64</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP32">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="704" name="data_add_2453724542" type="Const" version="opset1">
+ <data element_type="f32" offset="2226396" shape="1, 256, 1, 1" size="1024"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="705" name="bottleneck4_11/dim_inc/bn/variance/Fused_Add_" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_11/dim_inc/conv" precision="FP32">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="706" name="bottleneck4_11/add" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bottleneck4_11/add" precision="FP32">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="707" name="bottleneck4_11/fn/weights3094440475" type="Const" version="opset1">
+ <data element_type="f32" offset="4664" shape="1" size="4"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="708" name="bottleneck4_11/fn" type="PReLU" version="opset1">
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="bb_16xout_pd" precision="FP32">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="709" name="1046" type="Const" version="opset1">
+ <data element_type="f32" offset="2227420" shape="48, 256, 3, 3" size="442368"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>48</dim>
+ <dim>256</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="710" name="mbox_loc1/out/conv/WithoutBiases" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="1, 1" pads_end="1, 1" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>48</dim>
+ <dim>256</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP32">
+ <dim>1</dim>
+ <dim>48</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="711" name="mbox_loc1/out/conv/Dims13831" type="Const" version="opset1">
+ <data element_type="f32" offset="2669788" shape="1, 48, 1, 1" size="192"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>48</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="712" name="mbox_loc1/out/conv" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>48</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ <dim>48</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="mbox_loc1/out/conv" precision="FP32">
+ <dim>1</dim>
+ <dim>48</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="713" name="1296" type="Const" version="opset1">
+ <data element_type="i64" offset="2669980" shape="4" size="32"/>
+ <output>
+ <port id="0" precision="I64">
+ <dim>4</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="714" name="mbox_loc1/out/conv/perm" type="Transpose" version="opset1">
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>48</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="I64">
+ <dim>4</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="mbox_loc1/out/conv/perm" precision="FP32">
+ <dim>1</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ <dim>48</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="715" name="1308/shapes_concat" type="Const" version="opset1">
+ <data element_type="i64" offset="2670012" shape="2" size="16"/>
+ <output>
+ <port id="0" precision="I64">
+ <dim>2</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="716" name="mbox_loc1/out/conv/flat" type="Reshape" version="opset1">
+ <data special_zero="true"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ <dim>48</dim>
+ </port>
+ <port id="1" precision="I64">
+ <dim>2</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="mbox_loc1/out/conv/flat" precision="FP32">
+ <dim>1</dim>
+ <dim>32640</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="717" name="1004" type="Const" version="opset1">
+ <data element_type="f32" offset="2670028" shape="24, 256, 3, 3" size="221184"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>24</dim>
+ <dim>256</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="718" name="mbox_conf1/out/conv/WithoutBiases" type="Convolution" version="opset1">
+ <data auto_pad="explicit" dilations="1, 1" pads_begin="1, 1" pads_end="1, 1" strides="1, 1"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>24</dim>
+ <dim>256</dim>
+ <dim>3</dim>
+ <dim>3</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP32">
+ <dim>1</dim>
+ <dim>24</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="719" name="mbox_conf1/out/conv/Dims13825" type="Const" version="opset1">
+ <data element_type="f32" offset="2891212" shape="1, 24, 1, 1" size="96"/>
+ <output>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>24</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="720" name="mbox_conf1/out/conv" type="Add" version="opset1">
+ <data auto_broadcast="numpy"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>24</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ <dim>24</dim>
+ <dim>1</dim>
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="mbox_conf1/out/conv" precision="FP32">
+ <dim>1</dim>
+ <dim>24</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="721" name="1297" type="Const" version="opset1">
+ <data element_type="i64" offset="2669980" shape="4" size="32"/>
+ <output>
+ <port id="0" precision="I64">
+ <dim>4</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="722" name="mbox_conf1/out/conv/perm" type="Transpose" version="opset1">
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>24</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ <port id="1" precision="I64">
+ <dim>4</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="mbox_conf1/out/conv/perm" precision="FP32">
+ <dim>1</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ <dim>24</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="723" name="1303/shapes_concat" type="Const" version="opset1">
+ <data element_type="i64" offset="2670012" shape="2" size="16"/>
+ <output>
+ <port id="0" precision="I64">
+ <dim>2</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="724" name="mbox_conf1/out/conv/flat" type="Reshape" version="opset1">
+ <data special_zero="true"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ <dim>24</dim>
+ </port>
+ <port id="1" precision="I64">
+ <dim>2</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="mbox_conf1/out/conv/flat" precision="FP32">
+ <dim>1</dim>
+ <dim>16320</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="725" name="1295" type="Const" version="opset1">
+ <data element_type="i64" offset="2891308" shape="3" size="24"/>
+ <output>
+ <port id="0" precision="I64">
+ <dim>3</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="726" name="mbox_conf1/out/conv/flat/reshape" type="Reshape" version="opset1">
+ <data special_zero="true"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>16320</dim>
+ </port>
+ <port id="1" precision="I64">
+ <dim>3</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="mbox_conf1/out/conv/flat/reshape" precision="FP32">
+ <dim>1</dim>
+ <dim>8160</dim>
+ <dim>2</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="727" name="mbox_conf1/out/conv/flat/softmax" type="SoftMax" version="opset1">
+ <data axis="2"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>8160</dim>
+ <dim>2</dim>
+ </port>
+ </input>
+ <output>
+ <port id="1" names="mbox_conf1/out/conv/flat/softmax" precision="FP32">
+ <dim>1</dim>
+ <dim>8160</dim>
+ <dim>2</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="728" name="1298/shapes_concat" type="Const" version="opset1">
+ <data element_type="i64" offset="2670012" shape="2" size="16"/>
+ <output>
+ <port id="0" precision="I64">
+ <dim>2</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="729" name="mbox_conf1/out/conv/flat/softmax/flat" type="Reshape" version="opset1">
+ <data special_zero="true"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>8160</dim>
+ <dim>2</dim>
+ </port>
+ <port id="1" precision="I64">
+ <dim>2</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="mbox_conf1/out/conv/flat/softmax/flat" precision="FP32">
+ <dim>1</dim>
+ <dim>16320</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="730" name="mbox1/priorbox/0_port" type="ShapeOf" version="opset3">
+ <data output_type="i64"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>256</dim>
+ <dim>20</dim>
+ <dim>34</dim>
+ </port>
+ </input>
+ <output>
+ <port id="1" precision="I64">
+ <dim>4</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="731" name="mbox1/priorbox/ss_begin2978640640" type="Const" version="opset1">
+ <data element_type="i64" offset="2891332" shape="1" size="8"/>
+ <output>
+ <port id="0" precision="I64">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="732" name="mbox1/priorbox/ss_end2978740277" type="Const" version="opset1">
+ <data element_type="i64" offset="2891340" shape="1" size="8"/>
+ <output>
+ <port id="0" precision="I64">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="733" name="mbox1/priorbox/ss_stride2978839821" type="Const" version="opset1">
+ <data element_type="i64" offset="2891348" shape="1" size="8"/>
+ <output>
+ <port id="0" precision="I64">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="734" name="mbox1/priorbox/ss_0_port" type="StridedSlice" version="opset1">
+ <data begin_mask="0" ellipsis_mask="0" end_mask="1" new_axis_mask="0" shrink_axis_mask="0"/>
+ <input>
+ <port id="0" precision="I64">
+ <dim>4</dim>
+ </port>
+ <port id="1" precision="I64">
+ <dim>1</dim>
+ </port>
+ <port id="2" precision="I64">
+ <dim>1</dim>
+ </port>
+ <port id="3" precision="I64">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="4" precision="I64">
+ <dim>2</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="735" name="mbox1/priorbox/1_port" type="ShapeOf" version="opset3">
+ <data output_type="i64"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>3</dim>
+ <dim>320</dim>
+ <dim>544</dim>
+ </port>
+ </input>
+ <output>
+ <port id="1" precision="I64">
+ <dim>4</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="736" name="mbox1/priorbox/ss_begin2978639851" type="Const" version="opset1">
+ <data element_type="i64" offset="2891332" shape="1" size="8"/>
+ <output>
+ <port id="0" precision="I64">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="737" name="mbox1/priorbox/ss_end2978739902" type="Const" version="opset1">
+ <data element_type="i64" offset="2891340" shape="1" size="8"/>
+ <output>
+ <port id="0" precision="I64">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="738" name="mbox1/priorbox/ss_stride2978839968" type="Const" version="opset1">
+ <data element_type="i64" offset="2891348" shape="1" size="8"/>
+ <output>
+ <port id="0" precision="I64">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="739" name="mbox1/priorbox/ss_1_port" type="StridedSlice" version="opset1">
+ <data begin_mask="0" ellipsis_mask="0" end_mask="1" new_axis_mask="0" shrink_axis_mask="0"/>
+ <input>
+ <port id="0" precision="I64">
+ <dim>4</dim>
+ </port>
+ <port id="1" precision="I64">
+ <dim>1</dim>
+ </port>
+ <port id="2" precision="I64">
+ <dim>1</dim>
+ </port>
+ <port id="3" precision="I64">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="4" precision="I64">
+ <dim>2</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="740" name="mbox1/priorbox/naked_not_unsqueezed" type="PriorBoxClustered" version="opset1">
+ <data clip="false" height="34.07, 47.11, 54.22, 65.78, 75.56, 80.89, 89.78, 99.26, 115.56, 163.26, 194.07, 197.33" offset="0.5" step="0" step_h="16" step_w="16" variance="0.1, 0.1, 0.2, 0.2" width="11.33, 17, 20.68, 23.52, 28.05, 37.4, 30.03, 35.7, 44.2, 55.25, 78.12, 135.15"/>
+ <input>
+ <port id="0" precision="I64">
+ <dim>2</dim>
+ </port>
+ <port id="1" precision="I64">
+ <dim>2</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" precision="FP32">
+ <dim>2</dim>
+ <dim>32640</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="741" name="mbox1/priorbox/unsqueeze/value2979640376" type="Const" version="opset1">
+ <data element_type="i64" offset="2891356" shape="1" size="8"/>
+ <output>
+ <port id="0" precision="I64">
+ <dim>1</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="742" name="mbox1/priorbox" type="Unsqueeze" version="opset1">
+ <input>
+ <port id="0" precision="FP32">
+ <dim>2</dim>
+ <dim>32640</dim>
+ </port>
+ <port id="1" precision="I64">
+ <dim>1</dim>
+ </port>
+ </input>
+ <output>
+ <port id="2" names="mbox1/priorbox" precision="FP32">
+ <dim>1</dim>
+ <dim>2</dim>
+ <dim>32640</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="743" name="detection_out" type="DetectionOutput" version="opset1">
+ <data background_label_id="0" clip_after_nms="false" clip_before_nms="false" code_type="caffe.PriorBoxParameter.CENTER_SIZE" confidence_threshold="0.0099999997764825821" decrease_label_id="false" input_height="1" input_width="1" keep_top_k="200" nms_threshold="0.44999998807907104" normalized="true" num_classes="2" objectness_score="0" share_location="true" top_k="400" variance_encoded_in_target="false"/>
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>32640</dim>
+ </port>
+ <port id="1" precision="FP32">
+ <dim>1</dim>
+ <dim>16320</dim>
+ </port>
+ <port id="2" precision="FP32">
+ <dim>1</dim>
+ <dim>2</dim>
+ <dim>32640</dim>
+ </port>
+ </input>
+ <output>
+ <port id="3" names="detection_out" precision="FP32">
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>200</dim>
+ <dim>7</dim>
+ </port>
+ </output>
+ </layer>
+ <layer id="744" name="detection_out/sink_port_0" type="Result" version="opset1">
+ <input>
+ <port id="0" precision="FP32">
+ <dim>1</dim>
+ <dim>1</dim>
+ <dim>200</dim>
+ <dim>7</dim>
+ </port>
+ </input>
+ </layer>
+ </layers>
+ <edges>
+ <edge from-layer="0" from-port="0" to-layer="735" to-port="0"/>
+ <edge from-layer="0" from-port="0" to-layer="2" to-port="0"/>
+ <edge from-layer="1" from-port="0" to-layer="2" to-port="1"/>
+ <edge from-layer="2" from-port="2" to-layer="4" to-port="0"/>
+ <edge from-layer="3" from-port="0" to-layer="4" to-port="1"/>
+ <edge from-layer="4" from-port="2" to-layer="6" to-port="0"/>
+ <edge from-layer="5" from-port="0" to-layer="6" to-port="1"/>
+ <edge from-layer="6" from-port="2" to-layer="8" to-port="0"/>
+ <edge from-layer="7" from-port="0" to-layer="8" to-port="1"/>
+ <edge from-layer="8" from-port="2" to-layer="9" to-port="0"/>
+ <edge from-layer="9" from-port="1" to-layer="11" to-port="0"/>
+ <edge from-layer="9" from-port="1" to-layer="26" to-port="0"/>
+ <edge from-layer="10" from-port="0" to-layer="11" to-port="1"/>
+ <edge from-layer="11" from-port="2" to-layer="13" to-port="0"/>
+ <edge from-layer="12" from-port="0" to-layer="13" to-port="1"/>
+ <edge from-layer="13" from-port="2" to-layer="15" to-port="0"/>
+ <edge from-layer="14" from-port="0" to-layer="15" to-port="1"/>
+ <edge from-layer="15" from-port="2" to-layer="17" to-port="0"/>
+ <edge from-layer="16" from-port="0" to-layer="17" to-port="1"/>
+ <edge from-layer="17" from-port="2" to-layer="19" to-port="0"/>
+ <edge from-layer="18" from-port="0" to-layer="19" to-port="1"/>
+ <edge from-layer="19" from-port="2" to-layer="21" to-port="0"/>
+ <edge from-layer="20" from-port="0" to-layer="21" to-port="1"/>
+ <edge from-layer="21" from-port="2" to-layer="23" to-port="0"/>
+ <edge from-layer="22" from-port="0" to-layer="23" to-port="1"/>
+ <edge from-layer="23" from-port="2" to-layer="25" to-port="0"/>
+ <edge from-layer="24" from-port="0" to-layer="25" to-port="1"/>
+ <edge from-layer="25" from-port="2" to-layer="26" to-port="1"/>
+ <edge from-layer="26" from-port="2" to-layer="28" to-port="0"/>
+ <edge from-layer="27" from-port="0" to-layer="28" to-port="1"/>
+ <edge from-layer="28" from-port="2" to-layer="30" to-port="0"/>
+ <edge from-layer="28" from-port="2" to-layer="45" to-port="0"/>
+ <edge from-layer="29" from-port="0" to-layer="30" to-port="1"/>
+ <edge from-layer="30" from-port="2" to-layer="32" to-port="0"/>
+ <edge from-layer="31" from-port="0" to-layer="32" to-port="1"/>
+ <edge from-layer="32" from-port="2" to-layer="34" to-port="0"/>
+ <edge from-layer="33" from-port="0" to-layer="34" to-port="1"/>
+ <edge from-layer="34" from-port="2" to-layer="36" to-port="0"/>
+ <edge from-layer="35" from-port="0" to-layer="36" to-port="1"/>
+ <edge from-layer="36" from-port="2" to-layer="38" to-port="0"/>
+ <edge from-layer="37" from-port="0" to-layer="38" to-port="1"/>
+ <edge from-layer="38" from-port="2" to-layer="40" to-port="0"/>
+ <edge from-layer="39" from-port="0" to-layer="40" to-port="1"/>
+ <edge from-layer="40" from-port="2" to-layer="42" to-port="0"/>
+ <edge from-layer="41" from-port="0" to-layer="42" to-port="1"/>
+ <edge from-layer="42" from-port="2" to-layer="44" to-port="0"/>
+ <edge from-layer="43" from-port="0" to-layer="44" to-port="1"/>
+ <edge from-layer="44" from-port="2" to-layer="45" to-port="1"/>
+ <edge from-layer="45" from-port="2" to-layer="47" to-port="0"/>
+ <edge from-layer="46" from-port="0" to-layer="47" to-port="1"/>
+ <edge from-layer="47" from-port="2" to-layer="49" to-port="0"/>
+ <edge from-layer="47" from-port="2" to-layer="64" to-port="0"/>
+ <edge from-layer="48" from-port="0" to-layer="49" to-port="1"/>
+ <edge from-layer="49" from-port="2" to-layer="51" to-port="0"/>
+ <edge from-layer="50" from-port="0" to-layer="51" to-port="1"/>
+ <edge from-layer="51" from-port="2" to-layer="53" to-port="0"/>
+ <edge from-layer="52" from-port="0" to-layer="53" to-port="1"/>
+ <edge from-layer="53" from-port="2" to-layer="55" to-port="0"/>
+ <edge from-layer="54" from-port="0" to-layer="55" to-port="1"/>
+ <edge from-layer="55" from-port="2" to-layer="57" to-port="0"/>
+ <edge from-layer="56" from-port="0" to-layer="57" to-port="1"/>
+ <edge from-layer="57" from-port="2" to-layer="59" to-port="0"/>
+ <edge from-layer="58" from-port="0" to-layer="59" to-port="1"/>
+ <edge from-layer="59" from-port="2" to-layer="61" to-port="0"/>
+ <edge from-layer="60" from-port="0" to-layer="61" to-port="1"/>
+ <edge from-layer="61" from-port="2" to-layer="63" to-port="0"/>
+ <edge from-layer="62" from-port="0" to-layer="63" to-port="1"/>
+ <edge from-layer="63" from-port="2" to-layer="64" to-port="1"/>
+ <edge from-layer="64" from-port="2" to-layer="66" to-port="0"/>
+ <edge from-layer="65" from-port="0" to-layer="66" to-port="1"/>
+ <edge from-layer="66" from-port="2" to-layer="68" to-port="0"/>
+ <edge from-layer="66" from-port="2" to-layer="83" to-port="0"/>
+ <edge from-layer="67" from-port="0" to-layer="68" to-port="1"/>
+ <edge from-layer="68" from-port="2" to-layer="70" to-port="0"/>
+ <edge from-layer="69" from-port="0" to-layer="70" to-port="1"/>
+ <edge from-layer="70" from-port="2" to-layer="72" to-port="0"/>
+ <edge from-layer="71" from-port="0" to-layer="72" to-port="1"/>
+ <edge from-layer="72" from-port="2" to-layer="74" to-port="0"/>
+ <edge from-layer="73" from-port="0" to-layer="74" to-port="1"/>
+ <edge from-layer="74" from-port="2" to-layer="76" to-port="0"/>
+ <edge from-layer="75" from-port="0" to-layer="76" to-port="1"/>
+ <edge from-layer="76" from-port="2" to-layer="78" to-port="0"/>
+ <edge from-layer="77" from-port="0" to-layer="78" to-port="1"/>
+ <edge from-layer="78" from-port="2" to-layer="80" to-port="0"/>
+ <edge from-layer="79" from-port="0" to-layer="80" to-port="1"/>
+ <edge from-layer="80" from-port="2" to-layer="82" to-port="0"/>
+ <edge from-layer="81" from-port="0" to-layer="82" to-port="1"/>
+ <edge from-layer="82" from-port="2" to-layer="83" to-port="1"/>
+ <edge from-layer="83" from-port="2" to-layer="85" to-port="0"/>
+ <edge from-layer="84" from-port="0" to-layer="85" to-port="1"/>
+ <edge from-layer="85" from-port="2" to-layer="86" to-port="0"/>
+ <edge from-layer="85" from-port="2" to-layer="92" to-port="0"/>
+ <edge from-layer="86" from-port="1" to-layer="88" to-port="0"/>
+ <edge from-layer="87" from-port="0" to-layer="88" to-port="1"/>
+ <edge from-layer="88" from-port="2" to-layer="90" to-port="0"/>
+ <edge from-layer="89" from-port="0" to-layer="90" to-port="1"/>
+ <edge from-layer="90" from-port="2" to-layer="107" to-port="0"/>
+ <edge from-layer="91" from-port="0" to-layer="92" to-port="1"/>
+ <edge from-layer="92" from-port="2" to-layer="94" to-port="0"/>
+ <edge from-layer="93" from-port="0" to-layer="94" to-port="1"/>
+ <edge from-layer="94" from-port="2" to-layer="96" to-port="0"/>
+ <edge from-layer="95" from-port="0" to-layer="96" to-port="1"/>
+ <edge from-layer="96" from-port="2" to-layer="98" to-port="0"/>
+ <edge from-layer="97" from-port="0" to-layer="98" to-port="1"/>
+ <edge from-layer="98" from-port="2" to-layer="100" to-port="0"/>
+ <edge from-layer="99" from-port="0" to-layer="100" to-port="1"/>
+ <edge from-layer="100" from-port="2" to-layer="102" to-port="0"/>
+ <edge from-layer="101" from-port="0" to-layer="102" to-port="1"/>
+ <edge from-layer="102" from-port="2" to-layer="104" to-port="0"/>
+ <edge from-layer="103" from-port="0" to-layer="104" to-port="1"/>
+ <edge from-layer="104" from-port="2" to-layer="106" to-port="0"/>
+ <edge from-layer="105" from-port="0" to-layer="106" to-port="1"/>
+ <edge from-layer="106" from-port="2" to-layer="107" to-port="1"/>
+ <edge from-layer="107" from-port="2" to-layer="109" to-port="0"/>
+ <edge from-layer="108" from-port="0" to-layer="109" to-port="1"/>
+ <edge from-layer="109" from-port="2" to-layer="126" to-port="0"/>
+ <edge from-layer="109" from-port="2" to-layer="111" to-port="0"/>
+ <edge from-layer="110" from-port="0" to-layer="111" to-port="1"/>
+ <edge from-layer="111" from-port="2" to-layer="113" to-port="0"/>
+ <edge from-layer="112" from-port="0" to-layer="113" to-port="1"/>
+ <edge from-layer="113" from-port="2" to-layer="115" to-port="0"/>
+ <edge from-layer="114" from-port="0" to-layer="115" to-port="1"/>
+ <edge from-layer="115" from-port="2" to-layer="117" to-port="0"/>
+ <edge from-layer="116" from-port="0" to-layer="117" to-port="1"/>
+ <edge from-layer="117" from-port="2" to-layer="119" to-port="0"/>
+ <edge from-layer="118" from-port="0" to-layer="119" to-port="1"/>
+ <edge from-layer="119" from-port="2" to-layer="121" to-port="0"/>
+ <edge from-layer="120" from-port="0" to-layer="121" to-port="1"/>
+ <edge from-layer="121" from-port="2" to-layer="123" to-port="0"/>
+ <edge from-layer="122" from-port="0" to-layer="123" to-port="1"/>
+ <edge from-layer="123" from-port="2" to-layer="125" to-port="0"/>
+ <edge from-layer="124" from-port="0" to-layer="125" to-port="1"/>
+ <edge from-layer="125" from-port="2" to-layer="126" to-port="1"/>
+ <edge from-layer="126" from-port="2" to-layer="128" to-port="0"/>
+ <edge from-layer="127" from-port="0" to-layer="128" to-port="1"/>
+ <edge from-layer="128" from-port="2" to-layer="130" to-port="0"/>
+ <edge from-layer="128" from-port="2" to-layer="145" to-port="0"/>
+ <edge from-layer="129" from-port="0" to-layer="130" to-port="1"/>
+ <edge from-layer="130" from-port="2" to-layer="132" to-port="0"/>
+ <edge from-layer="131" from-port="0" to-layer="132" to-port="1"/>
+ <edge from-layer="132" from-port="2" to-layer="134" to-port="0"/>
+ <edge from-layer="133" from-port="0" to-layer="134" to-port="1"/>
+ <edge from-layer="134" from-port="2" to-layer="136" to-port="0"/>
+ <edge from-layer="135" from-port="0" to-layer="136" to-port="1"/>
+ <edge from-layer="136" from-port="2" to-layer="138" to-port="0"/>
+ <edge from-layer="137" from-port="0" to-layer="138" to-port="1"/>
+ <edge from-layer="138" from-port="2" to-layer="140" to-port="0"/>
+ <edge from-layer="139" from-port="0" to-layer="140" to-port="1"/>
+ <edge from-layer="140" from-port="2" to-layer="142" to-port="0"/>
+ <edge from-layer="141" from-port="0" to-layer="142" to-port="1"/>
+ <edge from-layer="142" from-port="2" to-layer="144" to-port="0"/>
+ <edge from-layer="143" from-port="0" to-layer="144" to-port="1"/>
+ <edge from-layer="144" from-port="2" to-layer="145" to-port="1"/>
+ <edge from-layer="145" from-port="2" to-layer="147" to-port="0"/>
+ <edge from-layer="146" from-port="0" to-layer="147" to-port="1"/>
+ <edge from-layer="147" from-port="2" to-layer="149" to-port="0"/>
+ <edge from-layer="147" from-port="2" to-layer="164" to-port="0"/>
+ <edge from-layer="148" from-port="0" to-layer="149" to-port="1"/>
+ <edge from-layer="149" from-port="2" to-layer="151" to-port="0"/>
+ <edge from-layer="150" from-port="0" to-layer="151" to-port="1"/>
+ <edge from-layer="151" from-port="2" to-layer="153" to-port="0"/>
+ <edge from-layer="152" from-port="0" to-layer="153" to-port="1"/>
+ <edge from-layer="153" from-port="2" to-layer="155" to-port="0"/>
+ <edge from-layer="154" from-port="0" to-layer="155" to-port="1"/>
+ <edge from-layer="155" from-port="2" to-layer="157" to-port="0"/>
+ <edge from-layer="156" from-port="0" to-layer="157" to-port="1"/>
+ <edge from-layer="157" from-port="2" to-layer="159" to-port="0"/>
+ <edge from-layer="158" from-port="0" to-layer="159" to-port="1"/>
+ <edge from-layer="159" from-port="2" to-layer="161" to-port="0"/>
+ <edge from-layer="160" from-port="0" to-layer="161" to-port="1"/>
+ <edge from-layer="161" from-port="2" to-layer="163" to-port="0"/>
+ <edge from-layer="162" from-port="0" to-layer="163" to-port="1"/>
+ <edge from-layer="163" from-port="2" to-layer="164" to-port="1"/>
+ <edge from-layer="164" from-port="2" to-layer="166" to-port="0"/>
+ <edge from-layer="165" from-port="0" to-layer="166" to-port="1"/>
+ <edge from-layer="166" from-port="2" to-layer="168" to-port="0"/>
+ <edge from-layer="166" from-port="2" to-layer="183" to-port="0"/>
+ <edge from-layer="167" from-port="0" to-layer="168" to-port="1"/>
+ <edge from-layer="168" from-port="2" to-layer="170" to-port="0"/>
+ <edge from-layer="169" from-port="0" to-layer="170" to-port="1"/>
+ <edge from-layer="170" from-port="2" to-layer="172" to-port="0"/>
+ <edge from-layer="171" from-port="0" to-layer="172" to-port="1"/>
+ <edge from-layer="172" from-port="2" to-layer="174" to-port="0"/>
+ <edge from-layer="173" from-port="0" to-layer="174" to-port="1"/>
+ <edge from-layer="174" from-port="2" to-layer="176" to-port="0"/>
+ <edge from-layer="175" from-port="0" to-layer="176" to-port="1"/>
+ <edge from-layer="176" from-port="2" to-layer="178" to-port="0"/>
+ <edge from-layer="177" from-port="0" to-layer="178" to-port="1"/>
+ <edge from-layer="178" from-port="2" to-layer="180" to-port="0"/>
+ <edge from-layer="179" from-port="0" to-layer="180" to-port="1"/>
+ <edge from-layer="180" from-port="2" to-layer="182" to-port="0"/>
+ <edge from-layer="181" from-port="0" to-layer="182" to-port="1"/>
+ <edge from-layer="182" from-port="2" to-layer="183" to-port="1"/>
+ <edge from-layer="183" from-port="2" to-layer="185" to-port="0"/>
+ <edge from-layer="184" from-port="0" to-layer="185" to-port="1"/>
+ <edge from-layer="185" from-port="2" to-layer="187" to-port="0"/>
+ <edge from-layer="185" from-port="2" to-layer="202" to-port="0"/>
+ <edge from-layer="186" from-port="0" to-layer="187" to-port="1"/>
+ <edge from-layer="187" from-port="2" to-layer="189" to-port="0"/>
+ <edge from-layer="188" from-port="0" to-layer="189" to-port="1"/>
+ <edge from-layer="189" from-port="2" to-layer="191" to-port="0"/>
+ <edge from-layer="190" from-port="0" to-layer="191" to-port="1"/>
+ <edge from-layer="191" from-port="2" to-layer="193" to-port="0"/>
+ <edge from-layer="192" from-port="0" to-layer="193" to-port="1"/>
+ <edge from-layer="193" from-port="2" to-layer="195" to-port="0"/>
+ <edge from-layer="194" from-port="0" to-layer="195" to-port="1"/>
+ <edge from-layer="195" from-port="2" to-layer="197" to-port="0"/>
+ <edge from-layer="196" from-port="0" to-layer="197" to-port="1"/>
+ <edge from-layer="197" from-port="2" to-layer="199" to-port="0"/>
+ <edge from-layer="198" from-port="0" to-layer="199" to-port="1"/>
+ <edge from-layer="199" from-port="2" to-layer="201" to-port="0"/>
+ <edge from-layer="200" from-port="0" to-layer="201" to-port="1"/>
+ <edge from-layer="201" from-port="2" to-layer="202" to-port="1"/>
+ <edge from-layer="202" from-port="2" to-layer="204" to-port="0"/>
+ <edge from-layer="203" from-port="0" to-layer="204" to-port="1"/>
+ <edge from-layer="204" from-port="2" to-layer="221" to-port="0"/>
+ <edge from-layer="204" from-port="2" to-layer="206" to-port="0"/>
+ <edge from-layer="205" from-port="0" to-layer="206" to-port="1"/>
+ <edge from-layer="206" from-port="2" to-layer="208" to-port="0"/>
+ <edge from-layer="207" from-port="0" to-layer="208" to-port="1"/>
+ <edge from-layer="208" from-port="2" to-layer="210" to-port="0"/>
+ <edge from-layer="209" from-port="0" to-layer="210" to-port="1"/>
+ <edge from-layer="210" from-port="2" to-layer="212" to-port="0"/>
+ <edge from-layer="211" from-port="0" to-layer="212" to-port="1"/>
+ <edge from-layer="212" from-port="2" to-layer="214" to-port="0"/>
+ <edge from-layer="213" from-port="0" to-layer="214" to-port="1"/>
+ <edge from-layer="214" from-port="2" to-layer="216" to-port="0"/>
+ <edge from-layer="215" from-port="0" to-layer="216" to-port="1"/>
+ <edge from-layer="216" from-port="2" to-layer="218" to-port="0"/>
+ <edge from-layer="217" from-port="0" to-layer="218" to-port="1"/>
+ <edge from-layer="218" from-port="2" to-layer="220" to-port="0"/>
+ <edge from-layer="219" from-port="0" to-layer="220" to-port="1"/>
+ <edge from-layer="220" from-port="2" to-layer="221" to-port="1"/>
+ <edge from-layer="221" from-port="2" to-layer="223" to-port="0"/>
+ <edge from-layer="222" from-port="0" to-layer="223" to-port="1"/>
+ <edge from-layer="223" from-port="2" to-layer="225" to-port="0"/>
+ <edge from-layer="223" from-port="2" to-layer="240" to-port="0"/>
+ <edge from-layer="224" from-port="0" to-layer="225" to-port="1"/>
+ <edge from-layer="225" from-port="2" to-layer="227" to-port="0"/>
+ <edge from-layer="226" from-port="0" to-layer="227" to-port="1"/>
+ <edge from-layer="227" from-port="2" to-layer="229" to-port="0"/>
+ <edge from-layer="228" from-port="0" to-layer="229" to-port="1"/>
+ <edge from-layer="229" from-port="2" to-layer="231" to-port="0"/>
+ <edge from-layer="230" from-port="0" to-layer="231" to-port="1"/>
+ <edge from-layer="231" from-port="2" to-layer="233" to-port="0"/>
+ <edge from-layer="232" from-port="0" to-layer="233" to-port="1"/>
+ <edge from-layer="233" from-port="2" to-layer="235" to-port="0"/>
+ <edge from-layer="234" from-port="0" to-layer="235" to-port="1"/>
+ <edge from-layer="235" from-port="2" to-layer="237" to-port="0"/>
+ <edge from-layer="236" from-port="0" to-layer="237" to-port="1"/>
+ <edge from-layer="237" from-port="2" to-layer="239" to-port="0"/>
+ <edge from-layer="238" from-port="0" to-layer="239" to-port="1"/>
+ <edge from-layer="239" from-port="2" to-layer="240" to-port="1"/>
+ <edge from-layer="240" from-port="2" to-layer="242" to-port="0"/>
+ <edge from-layer="241" from-port="0" to-layer="242" to-port="1"/>
+ <edge from-layer="242" from-port="2" to-layer="244" to-port="0"/>
+ <edge from-layer="242" from-port="2" to-layer="259" to-port="0"/>
+ <edge from-layer="243" from-port="0" to-layer="244" to-port="1"/>
+ <edge from-layer="244" from-port="2" to-layer="246" to-port="0"/>
+ <edge from-layer="245" from-port="0" to-layer="246" to-port="1"/>
+ <edge from-layer="246" from-port="2" to-layer="248" to-port="0"/>
+ <edge from-layer="247" from-port="0" to-layer="248" to-port="1"/>
+ <edge from-layer="248" from-port="2" to-layer="250" to-port="0"/>
+ <edge from-layer="249" from-port="0" to-layer="250" to-port="1"/>
+ <edge from-layer="250" from-port="2" to-layer="252" to-port="0"/>
+ <edge from-layer="251" from-port="0" to-layer="252" to-port="1"/>
+ <edge from-layer="252" from-port="2" to-layer="254" to-port="0"/>
+ <edge from-layer="253" from-port="0" to-layer="254" to-port="1"/>
+ <edge from-layer="254" from-port="2" to-layer="256" to-port="0"/>
+ <edge from-layer="255" from-port="0" to-layer="256" to-port="1"/>
+ <edge from-layer="256" from-port="2" to-layer="258" to-port="0"/>
+ <edge from-layer="257" from-port="0" to-layer="258" to-port="1"/>
+ <edge from-layer="258" from-port="2" to-layer="259" to-port="1"/>
+ <edge from-layer="259" from-port="2" to-layer="261" to-port="0"/>
+ <edge from-layer="260" from-port="0" to-layer="261" to-port="1"/>
+ <edge from-layer="261" from-port="2" to-layer="262" to-port="0"/>
+ <edge from-layer="261" from-port="2" to-layer="268" to-port="0"/>
+ <edge from-layer="262" from-port="1" to-layer="264" to-port="0"/>
+ <edge from-layer="263" from-port="0" to-layer="264" to-port="1"/>
+ <edge from-layer="264" from-port="2" to-layer="266" to-port="0"/>
+ <edge from-layer="265" from-port="0" to-layer="266" to-port="1"/>
+ <edge from-layer="266" from-port="2" to-layer="283" to-port="0"/>
+ <edge from-layer="267" from-port="0" to-layer="268" to-port="1"/>
+ <edge from-layer="268" from-port="2" to-layer="270" to-port="0"/>
+ <edge from-layer="269" from-port="0" to-layer="270" to-port="1"/>
+ <edge from-layer="270" from-port="2" to-layer="272" to-port="0"/>
+ <edge from-layer="271" from-port="0" to-layer="272" to-port="1"/>
+ <edge from-layer="272" from-port="2" to-layer="274" to-port="0"/>
+ <edge from-layer="273" from-port="0" to-layer="274" to-port="1"/>
+ <edge from-layer="274" from-port="2" to-layer="276" to-port="0"/>
+ <edge from-layer="275" from-port="0" to-layer="276" to-port="1"/>
+ <edge from-layer="276" from-port="2" to-layer="278" to-port="0"/>
+ <edge from-layer="277" from-port="0" to-layer="278" to-port="1"/>
+ <edge from-layer="278" from-port="2" to-layer="280" to-port="0"/>
+ <edge from-layer="279" from-port="0" to-layer="280" to-port="1"/>
+ <edge from-layer="280" from-port="2" to-layer="282" to-port="0"/>
+ <edge from-layer="281" from-port="0" to-layer="282" to-port="1"/>
+ <edge from-layer="282" from-port="2" to-layer="283" to-port="1"/>
+ <edge from-layer="283" from-port="2" to-layer="285" to-port="0"/>
+ <edge from-layer="284" from-port="0" to-layer="285" to-port="1"/>
+ <edge from-layer="285" from-port="2" to-layer="302" to-port="0"/>
+ <edge from-layer="285" from-port="2" to-layer="287" to-port="0"/>
+ <edge from-layer="286" from-port="0" to-layer="287" to-port="1"/>
+ <edge from-layer="287" from-port="2" to-layer="289" to-port="0"/>
+ <edge from-layer="288" from-port="0" to-layer="289" to-port="1"/>
+ <edge from-layer="289" from-port="2" to-layer="291" to-port="0"/>
+ <edge from-layer="290" from-port="0" to-layer="291" to-port="1"/>
+ <edge from-layer="291" from-port="2" to-layer="293" to-port="0"/>
+ <edge from-layer="292" from-port="0" to-layer="293" to-port="1"/>
+ <edge from-layer="293" from-port="2" to-layer="295" to-port="0"/>
+ <edge from-layer="294" from-port="0" to-layer="295" to-port="1"/>
+ <edge from-layer="295" from-port="2" to-layer="297" to-port="0"/>
+ <edge from-layer="296" from-port="0" to-layer="297" to-port="1"/>
+ <edge from-layer="297" from-port="2" to-layer="299" to-port="0"/>
+ <edge from-layer="298" from-port="0" to-layer="299" to-port="1"/>
+ <edge from-layer="299" from-port="2" to-layer="301" to-port="0"/>
+ <edge from-layer="300" from-port="0" to-layer="301" to-port="1"/>
+ <edge from-layer="301" from-port="2" to-layer="302" to-port="1"/>
+ <edge from-layer="302" from-port="2" to-layer="304" to-port="0"/>
+ <edge from-layer="303" from-port="0" to-layer="304" to-port="1"/>
+ <edge from-layer="304" from-port="2" to-layer="306" to-port="0"/>
+ <edge from-layer="304" from-port="2" to-layer="321" to-port="0"/>
+ <edge from-layer="305" from-port="0" to-layer="306" to-port="1"/>
+ <edge from-layer="306" from-port="2" to-layer="308" to-port="0"/>
+ <edge from-layer="307" from-port="0" to-layer="308" to-port="1"/>
+ <edge from-layer="308" from-port="2" to-layer="310" to-port="0"/>
+ <edge from-layer="309" from-port="0" to-layer="310" to-port="1"/>
+ <edge from-layer="310" from-port="2" to-layer="312" to-port="0"/>
+ <edge from-layer="311" from-port="0" to-layer="312" to-port="1"/>
+ <edge from-layer="312" from-port="2" to-layer="314" to-port="0"/>
+ <edge from-layer="313" from-port="0" to-layer="314" to-port="1"/>
+ <edge from-layer="314" from-port="2" to-layer="316" to-port="0"/>
+ <edge from-layer="315" from-port="0" to-layer="316" to-port="1"/>
+ <edge from-layer="316" from-port="2" to-layer="318" to-port="0"/>
+ <edge from-layer="317" from-port="0" to-layer="318" to-port="1"/>
+ <edge from-layer="318" from-port="2" to-layer="320" to-port="0"/>
+ <edge from-layer="319" from-port="0" to-layer="320" to-port="1"/>
+ <edge from-layer="320" from-port="2" to-layer="321" to-port="1"/>
+ <edge from-layer="321" from-port="2" to-layer="323" to-port="0"/>
+ <edge from-layer="322" from-port="0" to-layer="323" to-port="1"/>
+ <edge from-layer="323" from-port="2" to-layer="325" to-port="0"/>
+ <edge from-layer="323" from-port="2" to-layer="340" to-port="0"/>
+ <edge from-layer="324" from-port="0" to-layer="325" to-port="1"/>
+ <edge from-layer="325" from-port="2" to-layer="327" to-port="0"/>
+ <edge from-layer="326" from-port="0" to-layer="327" to-port="1"/>
+ <edge from-layer="327" from-port="2" to-layer="329" to-port="0"/>
+ <edge from-layer="328" from-port="0" to-layer="329" to-port="1"/>
+ <edge from-layer="329" from-port="2" to-layer="331" to-port="0"/>
+ <edge from-layer="330" from-port="0" to-layer="331" to-port="1"/>
+ <edge from-layer="331" from-port="2" to-layer="333" to-port="0"/>
+ <edge from-layer="332" from-port="0" to-layer="333" to-port="1"/>
+ <edge from-layer="333" from-port="2" to-layer="335" to-port="0"/>
+ <edge from-layer="334" from-port="0" to-layer="335" to-port="1"/>
+ <edge from-layer="335" from-port="2" to-layer="337" to-port="0"/>
+ <edge from-layer="336" from-port="0" to-layer="337" to-port="1"/>
+ <edge from-layer="337" from-port="2" to-layer="339" to-port="0"/>
+ <edge from-layer="338" from-port="0" to-layer="339" to-port="1"/>
+ <edge from-layer="339" from-port="2" to-layer="340" to-port="1"/>
+ <edge from-layer="340" from-port="2" to-layer="342" to-port="0"/>
+ <edge from-layer="341" from-port="0" to-layer="342" to-port="1"/>
+ <edge from-layer="342" from-port="2" to-layer="359" to-port="0"/>
+ <edge from-layer="342" from-port="2" to-layer="344" to-port="0"/>
+ <edge from-layer="343" from-port="0" to-layer="344" to-port="1"/>
+ <edge from-layer="344" from-port="2" to-layer="346" to-port="0"/>
+ <edge from-layer="345" from-port="0" to-layer="346" to-port="1"/>
+ <edge from-layer="346" from-port="2" to-layer="348" to-port="0"/>
+ <edge from-layer="347" from-port="0" to-layer="348" to-port="1"/>
+ <edge from-layer="348" from-port="2" to-layer="350" to-port="0"/>
+ <edge from-layer="349" from-port="0" to-layer="350" to-port="1"/>
+ <edge from-layer="350" from-port="2" to-layer="352" to-port="0"/>
+ <edge from-layer="351" from-port="0" to-layer="352" to-port="1"/>
+ <edge from-layer="352" from-port="2" to-layer="354" to-port="0"/>
+ <edge from-layer="353" from-port="0" to-layer="354" to-port="1"/>
+ <edge from-layer="354" from-port="2" to-layer="356" to-port="0"/>
+ <edge from-layer="355" from-port="0" to-layer="356" to-port="1"/>
+ <edge from-layer="356" from-port="2" to-layer="358" to-port="0"/>
+ <edge from-layer="357" from-port="0" to-layer="358" to-port="1"/>
+ <edge from-layer="358" from-port="2" to-layer="359" to-port="1"/>
+ <edge from-layer="359" from-port="2" to-layer="361" to-port="0"/>
+ <edge from-layer="360" from-port="0" to-layer="361" to-port="1"/>
+ <edge from-layer="361" from-port="2" to-layer="378" to-port="0"/>
+ <edge from-layer="361" from-port="2" to-layer="363" to-port="0"/>
+ <edge from-layer="362" from-port="0" to-layer="363" to-port="1"/>
+ <edge from-layer="363" from-port="2" to-layer="365" to-port="0"/>
+ <edge from-layer="364" from-port="0" to-layer="365" to-port="1"/>
+ <edge from-layer="365" from-port="2" to-layer="367" to-port="0"/>
+ <edge from-layer="366" from-port="0" to-layer="367" to-port="1"/>
+ <edge from-layer="367" from-port="2" to-layer="369" to-port="0"/>
+ <edge from-layer="368" from-port="0" to-layer="369" to-port="1"/>
+ <edge from-layer="369" from-port="2" to-layer="371" to-port="0"/>
+ <edge from-layer="370" from-port="0" to-layer="371" to-port="1"/>
+ <edge from-layer="371" from-port="2" to-layer="373" to-port="0"/>
+ <edge from-layer="372" from-port="0" to-layer="373" to-port="1"/>
+ <edge from-layer="373" from-port="2" to-layer="375" to-port="0"/>
+ <edge from-layer="374" from-port="0" to-layer="375" to-port="1"/>
+ <edge from-layer="375" from-port="2" to-layer="377" to-port="0"/>
+ <edge from-layer="376" from-port="0" to-layer="377" to-port="1"/>
+ <edge from-layer="377" from-port="2" to-layer="378" to-port="1"/>
+ <edge from-layer="378" from-port="2" to-layer="380" to-port="0"/>
+ <edge from-layer="379" from-port="0" to-layer="380" to-port="1"/>
+ <edge from-layer="380" from-port="2" to-layer="397" to-port="0"/>
+ <edge from-layer="380" from-port="2" to-layer="382" to-port="0"/>
+ <edge from-layer="381" from-port="0" to-layer="382" to-port="1"/>
+ <edge from-layer="382" from-port="2" to-layer="384" to-port="0"/>
+ <edge from-layer="383" from-port="0" to-layer="384" to-port="1"/>
+ <edge from-layer="384" from-port="2" to-layer="386" to-port="0"/>
+ <edge from-layer="385" from-port="0" to-layer="386" to-port="1"/>
+ <edge from-layer="386" from-port="2" to-layer="388" to-port="0"/>
+ <edge from-layer="387" from-port="0" to-layer="388" to-port="1"/>
+ <edge from-layer="388" from-port="2" to-layer="390" to-port="0"/>
+ <edge from-layer="389" from-port="0" to-layer="390" to-port="1"/>
+ <edge from-layer="390" from-port="2" to-layer="392" to-port="0"/>
+ <edge from-layer="391" from-port="0" to-layer="392" to-port="1"/>
+ <edge from-layer="392" from-port="2" to-layer="394" to-port="0"/>
+ <edge from-layer="393" from-port="0" to-layer="394" to-port="1"/>
+ <edge from-layer="394" from-port="2" to-layer="396" to-port="0"/>
+ <edge from-layer="395" from-port="0" to-layer="396" to-port="1"/>
+ <edge from-layer="396" from-port="2" to-layer="397" to-port="1"/>
+ <edge from-layer="397" from-port="2" to-layer="399" to-port="0"/>
+ <edge from-layer="398" from-port="0" to-layer="399" to-port="1"/>
+ <edge from-layer="399" from-port="2" to-layer="401" to-port="0"/>
+ <edge from-layer="399" from-port="2" to-layer="416" to-port="0"/>
+ <edge from-layer="400" from-port="0" to-layer="401" to-port="1"/>
+ <edge from-layer="401" from-port="2" to-layer="403" to-port="0"/>
+ <edge from-layer="402" from-port="0" to-layer="403" to-port="1"/>
+ <edge from-layer="403" from-port="2" to-layer="405" to-port="0"/>
+ <edge from-layer="404" from-port="0" to-layer="405" to-port="1"/>
+ <edge from-layer="405" from-port="2" to-layer="407" to-port="0"/>
+ <edge from-layer="406" from-port="0" to-layer="407" to-port="1"/>
+ <edge from-layer="407" from-port="2" to-layer="409" to-port="0"/>
+ <edge from-layer="408" from-port="0" to-layer="409" to-port="1"/>
+ <edge from-layer="409" from-port="2" to-layer="411" to-port="0"/>
+ <edge from-layer="410" from-port="0" to-layer="411" to-port="1"/>
+ <edge from-layer="411" from-port="2" to-layer="413" to-port="0"/>
+ <edge from-layer="412" from-port="0" to-layer="413" to-port="1"/>
+ <edge from-layer="413" from-port="2" to-layer="415" to-port="0"/>
+ <edge from-layer="414" from-port="0" to-layer="415" to-port="1"/>
+ <edge from-layer="415" from-port="2" to-layer="416" to-port="1"/>
+ <edge from-layer="416" from-port="2" to-layer="418" to-port="0"/>
+ <edge from-layer="417" from-port="0" to-layer="418" to-port="1"/>
+ <edge from-layer="418" from-port="2" to-layer="420" to-port="0"/>
+ <edge from-layer="418" from-port="2" to-layer="435" to-port="0"/>
+ <edge from-layer="419" from-port="0" to-layer="420" to-port="1"/>
+ <edge from-layer="420" from-port="2" to-layer="422" to-port="0"/>
+ <edge from-layer="421" from-port="0" to-layer="422" to-port="1"/>
+ <edge from-layer="422" from-port="2" to-layer="424" to-port="0"/>
+ <edge from-layer="423" from-port="0" to-layer="424" to-port="1"/>
+ <edge from-layer="424" from-port="2" to-layer="426" to-port="0"/>
+ <edge from-layer="425" from-port="0" to-layer="426" to-port="1"/>
+ <edge from-layer="426" from-port="2" to-layer="428" to-port="0"/>
+ <edge from-layer="427" from-port="0" to-layer="428" to-port="1"/>
+ <edge from-layer="428" from-port="2" to-layer="430" to-port="0"/>
+ <edge from-layer="429" from-port="0" to-layer="430" to-port="1"/>
+ <edge from-layer="430" from-port="2" to-layer="432" to-port="0"/>
+ <edge from-layer="431" from-port="0" to-layer="432" to-port="1"/>
+ <edge from-layer="432" from-port="2" to-layer="434" to-port="0"/>
+ <edge from-layer="433" from-port="0" to-layer="434" to-port="1"/>
+ <edge from-layer="434" from-port="2" to-layer="435" to-port="1"/>
+ <edge from-layer="435" from-port="2" to-layer="437" to-port="0"/>
+ <edge from-layer="436" from-port="0" to-layer="437" to-port="1"/>
+ <edge from-layer="437" from-port="2" to-layer="454" to-port="0"/>
+ <edge from-layer="437" from-port="2" to-layer="439" to-port="0"/>
+ <edge from-layer="438" from-port="0" to-layer="439" to-port="1"/>
+ <edge from-layer="439" from-port="2" to-layer="441" to-port="0"/>
+ <edge from-layer="440" from-port="0" to-layer="441" to-port="1"/>
+ <edge from-layer="441" from-port="2" to-layer="443" to-port="0"/>
+ <edge from-layer="442" from-port="0" to-layer="443" to-port="1"/>
+ <edge from-layer="443" from-port="2" to-layer="445" to-port="0"/>
+ <edge from-layer="444" from-port="0" to-layer="445" to-port="1"/>
+ <edge from-layer="445" from-port="2" to-layer="447" to-port="0"/>
+ <edge from-layer="446" from-port="0" to-layer="447" to-port="1"/>
+ <edge from-layer="447" from-port="2" to-layer="449" to-port="0"/>
+ <edge from-layer="448" from-port="0" to-layer="449" to-port="1"/>
+ <edge from-layer="449" from-port="2" to-layer="451" to-port="0"/>
+ <edge from-layer="450" from-port="0" to-layer="451" to-port="1"/>
+ <edge from-layer="451" from-port="2" to-layer="453" to-port="0"/>
+ <edge from-layer="452" from-port="0" to-layer="453" to-port="1"/>
+ <edge from-layer="453" from-port="2" to-layer="454" to-port="1"/>
+ <edge from-layer="454" from-port="2" to-layer="456" to-port="0"/>
+ <edge from-layer="455" from-port="0" to-layer="456" to-port="1"/>
+ <edge from-layer="456" from-port="2" to-layer="473" to-port="0"/>
+ <edge from-layer="456" from-port="2" to-layer="458" to-port="0"/>
+ <edge from-layer="457" from-port="0" to-layer="458" to-port="1"/>
+ <edge from-layer="458" from-port="2" to-layer="460" to-port="0"/>
+ <edge from-layer="459" from-port="0" to-layer="460" to-port="1"/>
+ <edge from-layer="460" from-port="2" to-layer="462" to-port="0"/>
+ <edge from-layer="461" from-port="0" to-layer="462" to-port="1"/>
+ <edge from-layer="462" from-port="2" to-layer="464" to-port="0"/>
+ <edge from-layer="463" from-port="0" to-layer="464" to-port="1"/>
+ <edge from-layer="464" from-port="2" to-layer="466" to-port="0"/>
+ <edge from-layer="465" from-port="0" to-layer="466" to-port="1"/>
+ <edge from-layer="466" from-port="2" to-layer="468" to-port="0"/>
+ <edge from-layer="467" from-port="0" to-layer="468" to-port="1"/>
+ <edge from-layer="468" from-port="2" to-layer="470" to-port="0"/>
+ <edge from-layer="469" from-port="0" to-layer="470" to-port="1"/>
+ <edge from-layer="470" from-port="2" to-layer="472" to-port="0"/>
+ <edge from-layer="471" from-port="0" to-layer="472" to-port="1"/>
+ <edge from-layer="472" from-port="2" to-layer="473" to-port="1"/>
+ <edge from-layer="473" from-port="2" to-layer="475" to-port="0"/>
+ <edge from-layer="474" from-port="0" to-layer="475" to-port="1"/>
+ <edge from-layer="475" from-port="2" to-layer="476" to-port="0"/>
+ <edge from-layer="475" from-port="2" to-layer="482" to-port="0"/>
+ <edge from-layer="476" from-port="1" to-layer="478" to-port="0"/>
+ <edge from-layer="477" from-port="0" to-layer="478" to-port="1"/>
+ <edge from-layer="478" from-port="2" to-layer="480" to-port="0"/>
+ <edge from-layer="479" from-port="0" to-layer="480" to-port="1"/>
+ <edge from-layer="480" from-port="2" to-layer="497" to-port="0"/>
+ <edge from-layer="481" from-port="0" to-layer="482" to-port="1"/>
+ <edge from-layer="482" from-port="2" to-layer="484" to-port="0"/>
+ <edge from-layer="483" from-port="0" to-layer="484" to-port="1"/>
+ <edge from-layer="484" from-port="2" to-layer="486" to-port="0"/>
+ <edge from-layer="485" from-port="0" to-layer="486" to-port="1"/>
+ <edge from-layer="486" from-port="2" to-layer="488" to-port="0"/>
+ <edge from-layer="487" from-port="0" to-layer="488" to-port="1"/>
+ <edge from-layer="488" from-port="2" to-layer="490" to-port="0"/>
+ <edge from-layer="489" from-port="0" to-layer="490" to-port="1"/>
+ <edge from-layer="490" from-port="2" to-layer="492" to-port="0"/>
+ <edge from-layer="491" from-port="0" to-layer="492" to-port="1"/>
+ <edge from-layer="492" from-port="2" to-layer="494" to-port="0"/>
+ <edge from-layer="493" from-port="0" to-layer="494" to-port="1"/>
+ <edge from-layer="494" from-port="2" to-layer="496" to-port="0"/>
+ <edge from-layer="495" from-port="0" to-layer="496" to-port="1"/>
+ <edge from-layer="496" from-port="2" to-layer="497" to-port="1"/>
+ <edge from-layer="497" from-port="2" to-layer="499" to-port="0"/>
+ <edge from-layer="498" from-port="0" to-layer="499" to-port="1"/>
+ <edge from-layer="499" from-port="2" to-layer="501" to-port="0"/>
+ <edge from-layer="499" from-port="2" to-layer="516" to-port="0"/>
+ <edge from-layer="500" from-port="0" to-layer="501" to-port="1"/>
+ <edge from-layer="501" from-port="2" to-layer="503" to-port="0"/>
+ <edge from-layer="502" from-port="0" to-layer="503" to-port="1"/>
+ <edge from-layer="503" from-port="2" to-layer="505" to-port="0"/>
+ <edge from-layer="504" from-port="0" to-layer="505" to-port="1"/>
+ <edge from-layer="505" from-port="2" to-layer="507" to-port="0"/>
+ <edge from-layer="506" from-port="0" to-layer="507" to-port="1"/>
+ <edge from-layer="507" from-port="2" to-layer="509" to-port="0"/>
+ <edge from-layer="508" from-port="0" to-layer="509" to-port="1"/>
+ <edge from-layer="509" from-port="2" to-layer="511" to-port="0"/>
+ <edge from-layer="510" from-port="0" to-layer="511" to-port="1"/>
+ <edge from-layer="511" from-port="2" to-layer="513" to-port="0"/>
+ <edge from-layer="512" from-port="0" to-layer="513" to-port="1"/>
+ <edge from-layer="513" from-port="2" to-layer="515" to-port="0"/>
+ <edge from-layer="514" from-port="0" to-layer="515" to-port="1"/>
+ <edge from-layer="515" from-port="2" to-layer="516" to-port="1"/>
+ <edge from-layer="516" from-port="2" to-layer="518" to-port="0"/>
+ <edge from-layer="517" from-port="0" to-layer="518" to-port="1"/>
+ <edge from-layer="518" from-port="2" to-layer="520" to-port="0"/>
+ <edge from-layer="518" from-port="2" to-layer="535" to-port="0"/>
+ <edge from-layer="519" from-port="0" to-layer="520" to-port="1"/>
+ <edge from-layer="520" from-port="2" to-layer="522" to-port="0"/>
+ <edge from-layer="521" from-port="0" to-layer="522" to-port="1"/>
+ <edge from-layer="522" from-port="2" to-layer="524" to-port="0"/>
+ <edge from-layer="523" from-port="0" to-layer="524" to-port="1"/>
+ <edge from-layer="524" from-port="2" to-layer="526" to-port="0"/>
+ <edge from-layer="525" from-port="0" to-layer="526" to-port="1"/>
+ <edge from-layer="526" from-port="2" to-layer="528" to-port="0"/>
+ <edge from-layer="527" from-port="0" to-layer="528" to-port="1"/>
+ <edge from-layer="528" from-port="2" to-layer="530" to-port="0"/>
+ <edge from-layer="529" from-port="0" to-layer="530" to-port="1"/>
+ <edge from-layer="530" from-port="2" to-layer="532" to-port="0"/>
+ <edge from-layer="531" from-port="0" to-layer="532" to-port="1"/>
+ <edge from-layer="532" from-port="2" to-layer="534" to-port="0"/>
+ <edge from-layer="533" from-port="0" to-layer="534" to-port="1"/>
+ <edge from-layer="534" from-port="2" to-layer="535" to-port="1"/>
+ <edge from-layer="535" from-port="2" to-layer="537" to-port="0"/>
+ <edge from-layer="536" from-port="0" to-layer="537" to-port="1"/>
+ <edge from-layer="537" from-port="2" to-layer="539" to-port="0"/>
+ <edge from-layer="537" from-port="2" to-layer="554" to-port="0"/>
+ <edge from-layer="538" from-port="0" to-layer="539" to-port="1"/>
+ <edge from-layer="539" from-port="2" to-layer="541" to-port="0"/>
+ <edge from-layer="540" from-port="0" to-layer="541" to-port="1"/>
+ <edge from-layer="541" from-port="2" to-layer="543" to-port="0"/>
+ <edge from-layer="542" from-port="0" to-layer="543" to-port="1"/>
+ <edge from-layer="543" from-port="2" to-layer="545" to-port="0"/>
+ <edge from-layer="544" from-port="0" to-layer="545" to-port="1"/>
+ <edge from-layer="545" from-port="2" to-layer="547" to-port="0"/>
+ <edge from-layer="546" from-port="0" to-layer="547" to-port="1"/>
+ <edge from-layer="547" from-port="2" to-layer="549" to-port="0"/>
+ <edge from-layer="548" from-port="0" to-layer="549" to-port="1"/>
+ <edge from-layer="549" from-port="2" to-layer="551" to-port="0"/>
+ <edge from-layer="550" from-port="0" to-layer="551" to-port="1"/>
+ <edge from-layer="551" from-port="2" to-layer="553" to-port="0"/>
+ <edge from-layer="552" from-port="0" to-layer="553" to-port="1"/>
+ <edge from-layer="553" from-port="2" to-layer="554" to-port="1"/>
+ <edge from-layer="554" from-port="2" to-layer="556" to-port="0"/>
+ <edge from-layer="555" from-port="0" to-layer="556" to-port="1"/>
+ <edge from-layer="556" from-port="2" to-layer="558" to-port="0"/>
+ <edge from-layer="556" from-port="2" to-layer="573" to-port="0"/>
+ <edge from-layer="557" from-port="0" to-layer="558" to-port="1"/>
+ <edge from-layer="558" from-port="2" to-layer="560" to-port="0"/>
+ <edge from-layer="559" from-port="0" to-layer="560" to-port="1"/>
+ <edge from-layer="560" from-port="2" to-layer="562" to-port="0"/>
+ <edge from-layer="561" from-port="0" to-layer="562" to-port="1"/>
+ <edge from-layer="562" from-port="2" to-layer="564" to-port="0"/>
+ <edge from-layer="563" from-port="0" to-layer="564" to-port="1"/>
+ <edge from-layer="564" from-port="2" to-layer="566" to-port="0"/>
+ <edge from-layer="565" from-port="0" to-layer="566" to-port="1"/>
+ <edge from-layer="566" from-port="2" to-layer="568" to-port="0"/>
+ <edge from-layer="567" from-port="0" to-layer="568" to-port="1"/>
+ <edge from-layer="568" from-port="2" to-layer="570" to-port="0"/>
+ <edge from-layer="569" from-port="0" to-layer="570" to-port="1"/>
+ <edge from-layer="570" from-port="2" to-layer="572" to-port="0"/>
+ <edge from-layer="571" from-port="0" to-layer="572" to-port="1"/>
+ <edge from-layer="572" from-port="2" to-layer="573" to-port="1"/>
+ <edge from-layer="573" from-port="2" to-layer="575" to-port="0"/>
+ <edge from-layer="574" from-port="0" to-layer="575" to-port="1"/>
+ <edge from-layer="575" from-port="2" to-layer="592" to-port="0"/>
+ <edge from-layer="575" from-port="2" to-layer="577" to-port="0"/>
+ <edge from-layer="576" from-port="0" to-layer="577" to-port="1"/>
+ <edge from-layer="577" from-port="2" to-layer="579" to-port="0"/>
+ <edge from-layer="578" from-port="0" to-layer="579" to-port="1"/>
+ <edge from-layer="579" from-port="2" to-layer="581" to-port="0"/>
+ <edge from-layer="580" from-port="0" to-layer="581" to-port="1"/>
+ <edge from-layer="581" from-port="2" to-layer="583" to-port="0"/>
+ <edge from-layer="582" from-port="0" to-layer="583" to-port="1"/>
+ <edge from-layer="583" from-port="2" to-layer="585" to-port="0"/>
+ <edge from-layer="584" from-port="0" to-layer="585" to-port="1"/>
+ <edge from-layer="585" from-port="2" to-layer="587" to-port="0"/>
+ <edge from-layer="586" from-port="0" to-layer="587" to-port="1"/>
+ <edge from-layer="587" from-port="2" to-layer="589" to-port="0"/>
+ <edge from-layer="588" from-port="0" to-layer="589" to-port="1"/>
+ <edge from-layer="589" from-port="2" to-layer="591" to-port="0"/>
+ <edge from-layer="590" from-port="0" to-layer="591" to-port="1"/>
+ <edge from-layer="591" from-port="2" to-layer="592" to-port="1"/>
+ <edge from-layer="592" from-port="2" to-layer="594" to-port="0"/>
+ <edge from-layer="593" from-port="0" to-layer="594" to-port="1"/>
+ <edge from-layer="594" from-port="2" to-layer="611" to-port="0"/>
+ <edge from-layer="594" from-port="2" to-layer="596" to-port="0"/>
+ <edge from-layer="595" from-port="0" to-layer="596" to-port="1"/>
+ <edge from-layer="596" from-port="2" to-layer="598" to-port="0"/>
+ <edge from-layer="597" from-port="0" to-layer="598" to-port="1"/>
+ <edge from-layer="598" from-port="2" to-layer="600" to-port="0"/>
+ <edge from-layer="599" from-port="0" to-layer="600" to-port="1"/>
+ <edge from-layer="600" from-port="2" to-layer="602" to-port="0"/>
+ <edge from-layer="601" from-port="0" to-layer="602" to-port="1"/>
+ <edge from-layer="602" from-port="2" to-layer="604" to-port="0"/>
+ <edge from-layer="603" from-port="0" to-layer="604" to-port="1"/>
+ <edge from-layer="604" from-port="2" to-layer="606" to-port="0"/>
+ <edge from-layer="605" from-port="0" to-layer="606" to-port="1"/>
+ <edge from-layer="606" from-port="2" to-layer="608" to-port="0"/>
+ <edge from-layer="607" from-port="0" to-layer="608" to-port="1"/>
+ <edge from-layer="608" from-port="2" to-layer="610" to-port="0"/>
+ <edge from-layer="609" from-port="0" to-layer="610" to-port="1"/>
+ <edge from-layer="610" from-port="2" to-layer="611" to-port="1"/>
+ <edge from-layer="611" from-port="2" to-layer="613" to-port="0"/>
+ <edge from-layer="612" from-port="0" to-layer="613" to-port="1"/>
+ <edge from-layer="613" from-port="2" to-layer="615" to-port="0"/>
+ <edge from-layer="613" from-port="2" to-layer="630" to-port="0"/>
+ <edge from-layer="614" from-port="0" to-layer="615" to-port="1"/>
+ <edge from-layer="615" from-port="2" to-layer="617" to-port="0"/>
+ <edge from-layer="616" from-port="0" to-layer="617" to-port="1"/>
+ <edge from-layer="617" from-port="2" to-layer="619" to-port="0"/>
+ <edge from-layer="618" from-port="0" to-layer="619" to-port="1"/>
+ <edge from-layer="619" from-port="2" to-layer="621" to-port="0"/>
+ <edge from-layer="620" from-port="0" to-layer="621" to-port="1"/>
+ <edge from-layer="621" from-port="2" to-layer="623" to-port="0"/>
+ <edge from-layer="622" from-port="0" to-layer="623" to-port="1"/>
+ <edge from-layer="623" from-port="2" to-layer="625" to-port="0"/>
+ <edge from-layer="624" from-port="0" to-layer="625" to-port="1"/>
+ <edge from-layer="625" from-port="2" to-layer="627" to-port="0"/>
+ <edge from-layer="626" from-port="0" to-layer="627" to-port="1"/>
+ <edge from-layer="627" from-port="2" to-layer="629" to-port="0"/>
+ <edge from-layer="628" from-port="0" to-layer="629" to-port="1"/>
+ <edge from-layer="629" from-port="2" to-layer="630" to-port="1"/>
+ <edge from-layer="630" from-port="2" to-layer="632" to-port="0"/>
+ <edge from-layer="631" from-port="0" to-layer="632" to-port="1"/>
+ <edge from-layer="632" from-port="2" to-layer="649" to-port="0"/>
+ <edge from-layer="632" from-port="2" to-layer="634" to-port="0"/>
+ <edge from-layer="633" from-port="0" to-layer="634" to-port="1"/>
+ <edge from-layer="634" from-port="2" to-layer="636" to-port="0"/>
+ <edge from-layer="635" from-port="0" to-layer="636" to-port="1"/>
+ <edge from-layer="636" from-port="2" to-layer="638" to-port="0"/>
+ <edge from-layer="637" from-port="0" to-layer="638" to-port="1"/>
+ <edge from-layer="638" from-port="2" to-layer="640" to-port="0"/>
+ <edge from-layer="639" from-port="0" to-layer="640" to-port="1"/>
+ <edge from-layer="640" from-port="2" to-layer="642" to-port="0"/>
+ <edge from-layer="641" from-port="0" to-layer="642" to-port="1"/>
+ <edge from-layer="642" from-port="2" to-layer="644" to-port="0"/>
+ <edge from-layer="643" from-port="0" to-layer="644" to-port="1"/>
+ <edge from-layer="644" from-port="2" to-layer="646" to-port="0"/>
+ <edge from-layer="645" from-port="0" to-layer="646" to-port="1"/>
+ <edge from-layer="646" from-port="2" to-layer="648" to-port="0"/>
+ <edge from-layer="647" from-port="0" to-layer="648" to-port="1"/>
+ <edge from-layer="648" from-port="2" to-layer="649" to-port="1"/>
+ <edge from-layer="649" from-port="2" to-layer="651" to-port="0"/>
+ <edge from-layer="650" from-port="0" to-layer="651" to-port="1"/>
+ <edge from-layer="651" from-port="2" to-layer="653" to-port="0"/>
+ <edge from-layer="651" from-port="2" to-layer="668" to-port="0"/>
+ <edge from-layer="652" from-port="0" to-layer="653" to-port="1"/>
+ <edge from-layer="653" from-port="2" to-layer="655" to-port="0"/>
+ <edge from-layer="654" from-port="0" to-layer="655" to-port="1"/>
+ <edge from-layer="655" from-port="2" to-layer="657" to-port="0"/>
+ <edge from-layer="656" from-port="0" to-layer="657" to-port="1"/>
+ <edge from-layer="657" from-port="2" to-layer="659" to-port="0"/>
+ <edge from-layer="658" from-port="0" to-layer="659" to-port="1"/>
+ <edge from-layer="659" from-port="2" to-layer="661" to-port="0"/>
+ <edge from-layer="660" from-port="0" to-layer="661" to-port="1"/>
+ <edge from-layer="661" from-port="2" to-layer="663" to-port="0"/>
+ <edge from-layer="662" from-port="0" to-layer="663" to-port="1"/>
+ <edge from-layer="663" from-port="2" to-layer="665" to-port="0"/>
+ <edge from-layer="664" from-port="0" to-layer="665" to-port="1"/>
+ <edge from-layer="665" from-port="2" to-layer="667" to-port="0"/>
+ <edge from-layer="666" from-port="0" to-layer="667" to-port="1"/>
+ <edge from-layer="667" from-port="2" to-layer="668" to-port="1"/>
+ <edge from-layer="668" from-port="2" to-layer="670" to-port="0"/>
+ <edge from-layer="669" from-port="0" to-layer="670" to-port="1"/>
+ <edge from-layer="670" from-port="2" to-layer="672" to-port="0"/>
+ <edge from-layer="670" from-port="2" to-layer="687" to-port="0"/>
+ <edge from-layer="671" from-port="0" to-layer="672" to-port="1"/>
+ <edge from-layer="672" from-port="2" to-layer="674" to-port="0"/>
+ <edge from-layer="673" from-port="0" to-layer="674" to-port="1"/>
+ <edge from-layer="674" from-port="2" to-layer="676" to-port="0"/>
+ <edge from-layer="675" from-port="0" to-layer="676" to-port="1"/>
+ <edge from-layer="676" from-port="2" to-layer="678" to-port="0"/>
+ <edge from-layer="677" from-port="0" to-layer="678" to-port="1"/>
+ <edge from-layer="678" from-port="2" to-layer="680" to-port="0"/>
+ <edge from-layer="679" from-port="0" to-layer="680" to-port="1"/>
+ <edge from-layer="680" from-port="2" to-layer="682" to-port="0"/>
+ <edge from-layer="681" from-port="0" to-layer="682" to-port="1"/>
+ <edge from-layer="682" from-port="2" to-layer="684" to-port="0"/>
+ <edge from-layer="683" from-port="0" to-layer="684" to-port="1"/>
+ <edge from-layer="684" from-port="2" to-layer="686" to-port="0"/>
+ <edge from-layer="685" from-port="0" to-layer="686" to-port="1"/>
+ <edge from-layer="686" from-port="2" to-layer="687" to-port="1"/>
+ <edge from-layer="687" from-port="2" to-layer="689" to-port="0"/>
+ <edge from-layer="688" from-port="0" to-layer="689" to-port="1"/>
+ <edge from-layer="689" from-port="2" to-layer="706" to-port="0"/>
+ <edge from-layer="689" from-port="2" to-layer="691" to-port="0"/>
+ <edge from-layer="690" from-port="0" to-layer="691" to-port="1"/>
+ <edge from-layer="691" from-port="2" to-layer="693" to-port="0"/>
+ <edge from-layer="692" from-port="0" to-layer="693" to-port="1"/>
+ <edge from-layer="693" from-port="2" to-layer="695" to-port="0"/>
+ <edge from-layer="694" from-port="0" to-layer="695" to-port="1"/>
+ <edge from-layer="695" from-port="2" to-layer="697" to-port="0"/>
+ <edge from-layer="696" from-port="0" to-layer="697" to-port="1"/>
+ <edge from-layer="697" from-port="2" to-layer="699" to-port="0"/>
+ <edge from-layer="698" from-port="0" to-layer="699" to-port="1"/>
+ <edge from-layer="699" from-port="2" to-layer="701" to-port="0"/>
+ <edge from-layer="700" from-port="0" to-layer="701" to-port="1"/>
+ <edge from-layer="701" from-port="2" to-layer="703" to-port="0"/>
+ <edge from-layer="702" from-port="0" to-layer="703" to-port="1"/>
+ <edge from-layer="703" from-port="2" to-layer="705" to-port="0"/>
+ <edge from-layer="704" from-port="0" to-layer="705" to-port="1"/>
+ <edge from-layer="705" from-port="2" to-layer="706" to-port="1"/>
+ <edge from-layer="706" from-port="2" to-layer="708" to-port="0"/>
+ <edge from-layer="707" from-port="0" to-layer="708" to-port="1"/>
+ <edge from-layer="708" from-port="2" to-layer="730" to-port="0"/>
+ <edge from-layer="708" from-port="2" to-layer="718" to-port="0"/>
+ <edge from-layer="708" from-port="2" to-layer="710" to-port="0"/>
+ <edge from-layer="709" from-port="0" to-layer="710" to-port="1"/>
+ <edge from-layer="710" from-port="2" to-layer="712" to-port="0"/>
+ <edge from-layer="711" from-port="0" to-layer="712" to-port="1"/>
+ <edge from-layer="712" from-port="2" to-layer="714" to-port="0"/>
+ <edge from-layer="713" from-port="0" to-layer="714" to-port="1"/>
+ <edge from-layer="714" from-port="2" to-layer="716" to-port="0"/>
+ <edge from-layer="715" from-port="0" to-layer="716" to-port="1"/>
+ <edge from-layer="716" from-port="2" to-layer="743" to-port="0"/>
+ <edge from-layer="717" from-port="0" to-layer="718" to-port="1"/>
+ <edge from-layer="718" from-port="2" to-layer="720" to-port="0"/>
+ <edge from-layer="719" from-port="0" to-layer="720" to-port="1"/>
+ <edge from-layer="720" from-port="2" to-layer="722" to-port="0"/>
+ <edge from-layer="721" from-port="0" to-layer="722" to-port="1"/>
+ <edge from-layer="722" from-port="2" to-layer="724" to-port="0"/>
+ <edge from-layer="723" from-port="0" to-layer="724" to-port="1"/>
+ <edge from-layer="724" from-port="2" to-layer="726" to-port="0"/>
+ <edge from-layer="725" from-port="0" to-layer="726" to-port="1"/>
+ <edge from-layer="726" from-port="2" to-layer="727" to-port="0"/>
+ <edge from-layer="727" from-port="1" to-layer="729" to-port="0"/>
+ <edge from-layer="728" from-port="0" to-layer="729" to-port="1"/>
+ <edge from-layer="729" from-port="2" to-layer="743" to-port="1"/>
+ <edge from-layer="730" from-port="1" to-layer="734" to-port="0"/>
+ <edge from-layer="731" from-port="0" to-layer="734" to-port="1"/>
+ <edge from-layer="732" from-port="0" to-layer="734" to-port="2"/>
+ <edge from-layer="733" from-port="0" to-layer="734" to-port="3"/>
+ <edge from-layer="734" from-port="4" to-layer="740" to-port="0"/>
+ <edge from-layer="735" from-port="1" to-layer="739" to-port="0"/>
+ <edge from-layer="736" from-port="0" to-layer="739" to-port="1"/>
+ <edge from-layer="737" from-port="0" to-layer="739" to-port="2"/>
+ <edge from-layer="738" from-port="0" to-layer="739" to-port="3"/>
+ <edge from-layer="739" from-port="4" to-layer="740" to-port="1"/>
+ <edge from-layer="740" from-port="2" to-layer="742" to-port="0"/>
+ <edge from-layer="741" from-port="0" to-layer="742" to-port="1"/>
+ <edge from-layer="742" from-port="2" to-layer="743" to-port="2"/>
+ <edge from-layer="743" from-port="3" to-layer="744" to-port="0"/>
+ </edges>
+ <meta_data>
+ <MO_version value="2021.4.0-3827-c5b65f2cb1d-releases/2021/4"/>
+ <cli_parameters>
+ <caffe_parser_path value="DIR"/>
+ <data_type value="FP32"/>
+ <disable_nhwc_to_nchw value="False"/>
+ <disable_omitting_optional value="False"/>
+ <disable_resnet_optimization value="False"/>
+ <disable_weights_compression value="False"/>
+ <enable_concat_optimization value="False"/>
+ <enable_flattening_nested_params value="False"/>
+ <enable_ssd_gluoncv value="False"/>
+ <extensions value="DIR"/>
+ <framework value="caffe"/>
+ <freeze_placeholder_with_value value="{}"/>
+ <generate_deprecated_IR_V7 value="False"/>
+ <input value="data"/>
+ <input_model value="DIR/rmnet_lrelu_pd_ssd.caffemodel"/>
+ <input_model_is_text value="False"/>
+ <input_proto value="DIR/rmnet_lrelu_pd_ssd.prototxt"/>
+ <input_shape value="[1,3,320,544]"/>
+ <k value="DIR/CustomLayersMapping.xml"/>
+ <keep_shape_ops value="True"/>
+ <legacy_ir_generation value="False"/>
+ <legacy_mxnet_model value="False"/>
+ <log_level value="ERROR"/>
+ <mean_scale_values value="{}"/>
+ <mean_values value="()"/>
+ <model_name value="person-detection-retail-0013"/>
+ <output value="['detection_out']"/>
+ <output_dir value="DIR"/>
+ <placeholder_data_types value="{}"/>
+ <placeholder_shapes value="{'data': array([ 1, 3, 320, 544])}"/>
+ <progress value="False"/>
+ <remove_memory value="False"/>
+ <remove_output_softmax value="False"/>
+ <reverse_input_channels value="False"/>
+ <save_params_from_nd value="False"/>
+ <scale_values value="()"/>
+ <silent value="False"/>
+ <static_shape value="False"/>
+ <stream_output value="False"/>
+ <transform value=""/>
+ <unset unset_cli_parameters="batch, counts, disable_fusing, disable_gfusing, finegrain_fusing, input_checkpoint, input_meta_graph, input_symbol, mean_file, mean_file_offsets, move_to_preprocess, nd_prefix_name, pretrained_model_name, saved_model_dir, saved_model_tags, scale, tensorboard_logdir, tensorflow_custom_layer_libraries, tensorflow_custom_operations_config_update, tensorflow_object_detection_api_pipeline_config, tensorflow_use_custom_operations_config, transformations_config"/>
+ </cli_parameters>
+ </meta_data>
+</net>
diff --git a/resources/run.mp4 b/resources/run.mp4
new file mode 100644
index 0000000..dd5f90e
--- /dev/null
+++ b/resources/run.mp4
Binary files differ