search

microsoft / MMdnn

MMdnn is a set of tools to help users inter-operate among different deep learning frameworks. E.g. model conversion and visualization. Convert models between Caffe, Keras, MXNet, Tensorflow, CNTK, PyTorch Onnx and CoreML.
MIT License
5.79k stars 964 forks source link

Uncaught exception when running the tutorial commands with Pytorch #468

Open charlesrwest opened 5 years ago

charlesrwest commented 5 years ago

Platform Ubuntu 18.04:

Python version: 3.6

Source framework with version: Pytorch 0.4.1

Destination framework with version: Caffe

Running the following commands (from https://github.com/Microsoft/MMdnn/blob/master/mmdnn/conversion/pytorch/README.md ) results in the error below:

Installed via pip3

Commands: mmdownload -f pytorch -n resnet101 -o ./ mmtoir -f pytorch -d resnet101 --inputShape 3,224,224 -n imagenet_resnet101.pth

Error Message:

Traceback (most recent call last): File "/usr/local/bin/mmtoir", line 11, in sys.exit(_main()) File "/usr/local/lib/python3.6/dist-packages/mmdnn/conversion/_script/convertToIR.py", line 192, in _main ret = _convert(args) File "/usr/local/lib/python3.6/dist-packages/mmdnn/conversion/_script/convertToIR.py", line 92, in _convert parser = PytorchParser(args.network, inputshape[0]) File "/usr/local/lib/python3.6/dist-packages/mmdnn/conversion/pytorch/pytorch_parser.py", line 83, in init self.pytorch_graph.build(self.input_shape) File "/usr/local/lib/python3.6/dist-packages/mmdnn/conversion/pytorch/pytorch_graph.py", line 124, in build trace.set_graph(PytorchGraph._optimize_graph(trace.graph(), False)) File "/usr/local/lib/python3.6/dist-packages/mmdnn/conversion/pytorch/pytorch_graph.py", line 74, in _optimize_graph graph = torch._C._jit_pass_onnx(graph, aten) TypeError: _jit_pass_onnx(): incompatible function arguments. The following argument types are supported:

  1. (arg0: torch::jit::Graph, arg1: torch._C._onnx.OperatorExportTypes) -> torch::jit::Graph

Invoked with: graph(%0 : Float(1, 3, 224, 224) %1 : Float(64, 3, 7, 7) %2 : Float(64) %3 : Float(64) %4 : Float(64) %5 : Float(64) %6 : Long() %7 : Float(64, 64, 1, 1) %8 : Float(64) %9 : Float(64) %10 : Float(64) %11 : Float(64) %12 : Long() %13 : Float(64, 64, 3, 3) %14 : Float(64) %15 : Float(64) %16 : Float(64) %17 : Float(64) %18 : Long() %19 : Float(256, 64, 1, 1) %20 : Float(256) %21 : Float(256) %22 : Float(256) %23 : Float(256) %24 : Long() %25 : Float(256, 64, 1, 1) %26 : Float(256) %27 : Float(256) %28 : Float(256) %29 : Float(256) %30 : Long() %31 : Float(64, 256, 1, 1) %32 : Float(64) %33 : Float(64) %34 : Float(64) %35 : Float(64) %36 : Long() %37 : Float(64, 64, 3, 3) %38 : Float(64) %39 : Float(64) %40 : Float(64) %41 : Float(64) %42 : Long() %43 : Float(256, 64, 1, 1) %44 : Float(256) %45 : Float(256) %46 : Float(256) %47 : Float(256) %48 : Long() %49 : Float(64, 256, 1, 1) %50 : Float(64) %51 : Float(64) %52 : Float(64) %53 : Float(64) %54 : Long() %55 : Float(64, 64, 3, 3) %56 : Float(64) %57 : Float(64) %58 : Float(64) %59 : Float(64) %60 : Long() %61 : Float(256, 64, 1, 1) %62 : Float(256) %63 : Float(256) %64 : Float(256) %65 : Float(256) %66 : Long() %67 : Float(128, 256, 1, 1) %68 : Float(128) %69 : Float(128) %70 : Float(128) %71 : Float(128) %72 : Long() %73 : Float(128, 128, 3, 3) %74 : Float(128) %75 : Float(128) %76 : Float(128) %77 : Float(128) %78 : Long() %79 : Float(512, 128, 1, 1) %80 : Float(512) %81 : Float(512) %82 : Float(512) %83 : Float(512) %84 : Long() %85 : Float(512, 256, 1, 1) %86 : Float(512) %87 : Float(512) %88 : Float(512) %89 : Float(512) %90 : Long() %91 : Float(128, 512, 1, 1) %92 : Float(128) %93 : Float(128) %94 : Float(128) %95 : Float(128) %96 : Long() %97 : Float(128, 128, 3, 3) %98 : Float(128) %99 : Float(128) %100 : Float(128) %101 : Float(128) %102 : Long() %103 : Float(512, 128, 1, 1) %104 : Float(512) %105 : Float(512) %106 : Float(512) %107 : Float(512) %108 : Long() %109 : Float(128, 512, 1, 1) %110 : Float(128) %111 : Float(128) %112 : Float(128) %113 : Float(128) %114 : Long() %115 : Float(128, 128, 3, 3) %116 : Float(128) %117 : Float(128) %118 : Float(128) %119 : Float(128) %120 : Long() %121 : Float(512, 128, 1, 1) %122 : Float(512) %123 : Float(512) %124 : Float(512) %125 : Float(512) %126 : Long() %127 : Float(128, 512, 1, 1) %128 : Float(128) %129 : Float(128) %130 : Float(128) %131 : Float(128) %132 : Long() %133 : Float(128, 128, 3, 3) %134 : Float(128) %135 : Float(128) %136 : Float(128) %137 : Float(128) %138 : Long() %139 : Float(512, 128, 1, 1) %140 : Float(512) %141 : Float(512) %142 : Float(512) %143 : Float(512) %144 : Long() %145 : Float(256, 512, 1, 1) %146 : Float(256) %147 : Float(256) %148 : Float(256) %149 : Float(256) %150 : Long() %151 : Float(256, 256, 3, 3) %152 : Float(256) %153 : Float(256) %154 : Float(256) %155 : Float(256) %156 : Long() %157 : Float(1024, 256, 1, 1) %158 : Float(1024) %159 : Float(1024) %160 : Float(1024) %161 : Float(1024) %162 : Long() %163 : Float(1024, 512, 1, 1) %164 : Float(1024) %165 : Float(1024) %166 : Float(1024) %167 : Float(1024) %168 : Long() %169 : Float(256, 1024, 1, 1) %170 : Float(256) %171 : Float(256) %172 : Float(256) %173 : Float(256) %174 : Long() %175 : Float(256, 256, 3, 3) %176 : Float(256) %177 : Float(256) %178 : Float(256) %179 : Float(256) %180 : Long() %181 : Float(1024, 256, 1, 1) %182 : Float(1024) %183 : Float(1024) %184 : Float(1024) %185 : Float(1024) %186 : Long() %187 : Float(256, 1024, 1, 1) %188 : Float(256) %189 : Float(256) %190 : Float(256) %191 : Float(256) %192 : Long() %193 : Float(256, 256, 3, 3) %194 : Float(256) %195 : Float(256) %196 : Float(256) %197 : Float(256) %198 : Long() %199 : Float(1024, 256, 1, 1) %200 : Float(1024) %201 : Float(1024) %202 : Float(1024) %203 : Float(1024) %204 : Long() %205 : Float(256, 1024, 1, 1) %206 : Float(256) %207 : Float(256) %208 : Float(256) %209 : Float(256) %210 : Long() %211 : Float(256, 256, 3, 3) %212 : Float(256) %213 : Float(256) %214 : Float(256) %215 : Float(256) %216 : Long() %217 : Float(1024, 256, 1, 1) %218 : Float(1024) %219 : Float(1024) %220 : Float(1024) %221 : Float(1024) %222 : Long() %223 : Float(256, 1024, 1, 1) %224 : Float(256) %225 : Float(256) %226 : Float(256) %227 : Float(256) %228 : Long() %229 : Float(256, 256, 3, 3) %230 : Float(256) %231 : Float(256) %232 : Float(256) %233 : Float(256) %234 : Long() %235 : Float(1024, 256, 1, 1) %236 : Float(1024) %237 : Float(1024) %238 : Float(1024) %239 : Float(1024) %240 : Long() %241 : Float(256, 1024, 1, 1) %242 : Float(256) %243 : Float(256) %244 : Float(256) %245 : Float(256) %246 : Long() %247 : Float(256, 256, 3, 3) %248 : Float(256) %249 : Float(256) %250 : Float(256) %251 : Float(256) %252 : Long() %253 : Float(1024, 256, 1, 1) %254 : Float(1024) %255 : Float(1024) %256 : Float(1024) %257 : Float(1024) %258 : Long() %259 : Float(256, 1024, 1, 1) %260 : Float(256) %261 : Float(256) %262 : Float(256) %263 : Float(256) %264 : Long() %265 : Float(256, 256, 3, 3) %266 : Float(256) %267 : Float(256) %268 : Float(256) %269 : Float(256) %270 : Long() %271 : Float(1024, 256, 1, 1) %272 : Float(1024) %273 : Float(1024) %274 : Float(1024) %275 : Float(1024) %276 : Long() %277 : Float(256, 1024, 1, 1) %278 : Float(256) %279 : Float(256) %280 : Float(256) %281 : Float(256) %282 : Long() %283 : Float(256, 256, 3, 3) %284 : Float(256) %285 : Float(256) %286 : Float(256) %287 : Float(256) %288 : Long() %289 : Float(1024, 256, 1, 1) %290 : Float(1024) %291 : Float(1024) %292 : Float(1024) %293 : Float(1024) %294 : Long() %295 : Float(256, 1024, 1, 1) %296 : Float(256) %297 : Float(256) %298 : Float(256) %299 : Float(256) %300 : Long() %301 : Float(256, 256, 3, 3) %302 : Float(256) %303 : Float(256) %304 : Float(256) %305 : Float(256) %306 : Long() %307 : Float(1024, 256, 1, 1) %308 : Float(1024) %309 : Float(1024) %310 : Float(1024) %311 : Float(1024) %312 : Long() %313 : Float(256, 1024, 1, 1) %314 : Float(256) %315 : Float(256) %316 : Float(256) %317 : Float(256) %318 : Long() %319 : Float(256, 256, 3, 3) %320 : Float(256) %321 : Float(256) %322 : Float(256) %323 : Float(256) %324 : Long() %325 : Float(1024, 256, 1, 1) %326 : Float(1024) %327 : Float(1024) %328 : Float(1024) %329 : Float(1024) %330 : Long() %331 : Float(256, 1024, 1, 1) %332 : Float(256) %333 : Float(256) %334 : Float(256) %335 : Float(256) %336 : Long() %337 : Float(256, 256, 3, 3) %338 : Float(256) %339 : Float(256) %340 : Float(256) %341 : Float(256) %342 : Long() %343 : Float(1024, 256, 1, 1) %344 : Float(1024) %345 : Float(1024) %346 : Float(1024) %347 : Float(1024) %348 : Long() %349 : Float(256, 1024, 1, 1) %350 : Float(256) %351 : Float(256) %352 : Float(256) %353 : Float(256) %354 : Long() %355 : Float(256, 256, 3, 3) %356 : Float(256) %357 : Float(256) %358 : Float(256) %359 : Float(256) %360 : Long() %361 : Float(1024, 256, 1, 1) %362 : Float(1024) %363 : Float(1024) %364 : Float(1024) %365 : Float(1024) %366 : Long() %367 : Float(256, 1024, 1, 1) %368 : Float(256) %369 : Float(256) %370 : Float(256) %371 : Float(256) %372 : Long() %373 : Float(256, 256, 3, 3) %374 : Float(256) %375 : Float(256) %376 : Float(256) %377 : Float(256) %378 : Long() %379 : Float(1024, 256, 1, 1) %380 : Float(1024) %381 : Float(1024) %382 : Float(1024) %383 : Float(1024) %384 : Long() %385 : Float(256, 1024, 1, 1) %386 : Float(256) %387 : Float(256) %388 : Float(256) %389 : Float(256) %390 : Long() %391 : Float(256, 256, 3, 3) %392 : Float(256) %393 : Float(256) %394 : Float(256) %395 : Float(256) %396 : Long() %397 : Float(1024, 256, 1, 1) %398 : Float(1024) %399 : Float(1024) %400 : Float(1024) %401 : Float(1024) %402 : Long() %403 : Float(256, 1024, 1, 1) %404 : Float(256) %405 : Float(256) %406 : Float(256) %407 : Float(256) %408 : Long() %409 : Float(256, 256, 3, 3) %410 : Float(256) %411 : Float(256) %412 : Float(256) %413 : Float(256) %414 : Long() %415 : Float(1024, 256, 1, 1) %416 : Float(1024) %417 : Float(1024) %418 : Float(1024) %419 : Float(1024) %420 : Long() %421 : Float(256, 1024, 1, 1) %422 : Float(256) %423 : Float(256) %424 : Float(256) %425 : Float(256) %426 : Long() %427 : Float(256, 256, 3, 3) %428 : Float(256) %429 : Float(256) %430 : Float(256) %431 : Float(256) %432 : Long() %433 : Float(1024, 256, 1, 1) %434 : Float(1024) %435 : Float(1024) %436 : Float(1024) %437 : Float(1024) %438 : Long() %439 : Float(256, 1024, 1, 1) %440 : Float(256) %441 : Float(256) %442 : Float(256) %443 : Float(256) %444 : Long() %445 : Float(256, 256, 3, 3) %446 : Float(256) %447 : Float(256) %448 : Float(256) %449 : Float(256) %450 : Long() %451 : Float(1024, 256, 1, 1) %452 : Float(1024) %453 : Float(1024) %454 : Float(1024) %455 : Float(1024) %456 : Long() %457 : Float(256, 1024, 1, 1) %458 : Float(256) %459 : Float(256) %460 : Float(256) %461 : Float(256) %462 : Long() %463 : Float(256, 256, 3, 3) %464 : Float(256) %465 : Float(256) %466 : Float(256) %467 : Float(256) %468 : Long() %469 : Float(1024, 256, 1, 1) %470 : Float(1024) %471 : Float(1024) %472 : Float(1024) %473 : Float(1024) %474 : Long() %475 : Float(256, 1024, 1, 1) %476 : Float(256) %477 : Float(256) %478 : Float(256) %479 : Float(256) %480 : Long() %481 : Float(256, 256, 3, 3) %482 : Float(256) %483 : Float(256) %484 : Float(256) %485 : Float(256) %486 : Long() %487 : Float(1024, 256, 1, 1) %488 : Float(1024) %489 : Float(1024) %490 : Float(1024) %491 : Float(1024) %492 : Long() %493 : Float(256, 1024, 1, 1) %494 : Float(256) %495 : Float(256) %496 : Float(256) %497 : Float(256) %498 : Long() %499 : Float(256, 256, 3, 3) %500 : Float(256) %501 : Float(256) %502 : Float(256) %503 : Float(256) %504 : Long() %505 : Float(1024, 256, 1, 1) %506 : Float(1024) %507 : Float(1024) %508 : Float(1024) %509 : Float(1024) %510 : Long() %511 : Float(256, 1024, 1, 1) %512 : Float(256) %513 : Float(256) %514 : Float(256) %515 : Float(256) %516 : Long() %517 : Float(256, 256, 3, 3) %518 : Float(256) %519 : Float(256) %520 : Float(256) %521 : Float(256) %522 : Long() %523 : Float(1024, 256, 1, 1) %524 : Float(1024) %525 : Float(1024) %526 : Float(1024) %527 : Float(1024) %528 : Long() %529 : Float(256, 1024, 1, 1) %530 : Float(256) %531 : Float(256) %532 : Float(256) %533 : Float(256) %534 : Long() %535 : Float(256, 256, 3, 3) %536 : Float(256) %537 : Float(256) %538 : Float(256) %539 : Float(256) %540 : Long() %541 : Float(1024, 256, 1, 1) %542 : Float(1024) %543 : Float(1024) %544 : Float(1024) %545 : Float(1024) %546 : Long() %547 : Float(256, 1024, 1, 1) %548 : Float(256) %549 : Float(256) %550 : Float(256) %551 : Float(256) %552 : Long() %553 : Float(256, 256, 3, 3) %554 : Float(256) %555 : Float(256) %556 : Float(256) %557 : Float(256) %558 : Long() %559 : Float(1024, 256, 1, 1) %560 : Float(1024) %561 : Float(1024) %562 : Float(1024) %563 : Float(1024) %564 : Long() %565 : Float(512, 1024, 1, 1) %566 : Float(512) %567 : Float(512) %568 : Float(512) %569 : Float(512) %570 : Long() %571 : Float(512, 512, 3, 3) %572 : Float(512) %573 : Float(512) %574 : Float(512) %575 : Float(512) %576 : Long() %577 : Float(2048, 512, 1, 1) %578 : Float(2048) %579 : Float(2048) %580 : Float(2048) %581 : Float(2048) %582 : Long() %583 : Float(2048, 1024, 1, 1) %584 : Float(2048) %585 : Float(2048) %586 : Float(2048) %587 : Float(2048) %588 : Long() %589 : Float(512, 2048, 1, 1) %590 : Float(512) %591 : Float(512) %592 : Float(512) %593 : Float(512) %594 : Long() %595 : Float(512, 512, 3, 3) %596 : Float(512) %597 : Float(512) %598 : Float(512) %599 : Float(512) %600 : Long() %601 : Float(2048, 512, 1, 1) %602 : Float(2048) %603 : Float(2048) %604 : Float(2048) %605 : Float(2048) %606 : Long() %607 : Float(512, 2048, 1, 1) %608 : Float(512) %609 : Float(512) %610 : Float(512) %611 : Float(512) %612 : Long() %613 : Float(512, 512, 3, 3) %614 : Float(512) %615 : Float(512) %616 : Float(512) %617 : Float(512) %618 : Long() %619 : Float(2048, 512, 1, 1) %620 : Float(2048) %621 : Float(2048) %622 : Float(2048) %623 : Float(2048) %624 : Long() %625 : Float(1000, 2048) %626 : Float(1000)) { %627 : Dynamic = prim::Undefined(), scope: ResNet/Conv2d[conv1] %636 : Float(1, 64, 112, 112) = aten::_convolution[stride=[2, 2], padding=[3, 3], dilation=[1, 1], transposed=0, output_padding=[0, 0], groups=1, benchmark=0, deterministic=0, cudnn_enabled=1](%0, %1, %627), scope: ResNet/Conv2d[conv1] %641 : Float(1, 64, 112, 112) = aten::batch_norm[training=0, momentum=0, eps=1e-05, cudnn_enabled=1](%636, %2, %3, %4, %5), scope: ResNet/BatchNorm2d[bn1] %643 : Float(1, 64, 112, 112) = aten::thresholdthreshold={0}, value={0}, scope: ResNet/ReLU[relu] %646 : Float(1, 64, 56, 56), %647 : Long(1, 64, 56, 56) = aten::max_pool2d_with_indiceskernel_size=[3, 3], stride=[2, 2], padding=[1, 1], dilation=[1, 1], ceil_mode=0, scope: ResNet/MaxPool2d[maxpool] %648 : Dynamic = prim::Undefined(), scope: ResNet/Sequential[layer1]/Bottleneck[0]/Conv2d[conv1] %657 : Float(1, 64, 56, 56) = aten::_convolution[stride=[1, 1], padding=[0, 0], dilation=[1, 1], transposed=0, output_padding=[0, 0], groups=1, benchmark=0, deterministic=0, cudnn_enabled=1](%646, %7, %648), scope: ResNet/Sequential[layer1]/Bottleneck[0]/Conv2d[conv1] %662 : Float(1, 64, 56, 56) = aten::batch_norm[training=0, momentum=0, eps=1e-05, cudnn_enabled=1](%657, %8, %9, %10, %11), scope: ResNet/Sequential[layer1]/Bottleneck[0]/BatchNorm2d[bn1] %664 : Float(1, 64, 56, 56) = aten::thresholdthreshold={0}, value={0}, scope: ResNet/Sequential[layer1]/Bottleneck[0]/ReLU[relu] %665 : Dynamic = prim::Undefined(), scope: ResNet/Sequential[layer1]/Bottleneck[0]/Conv2d[conv2] %674 : Float(1, 64, 56, 56) = aten::_convolution[stride=[1, 1], padding=[1, 1], dilation=[1, 1], transposed=0, output_padding=[0, 0], groups=1, benchmark=0, deterministic=0, cudnn_enabled=1](%664, %13, %665), scope: ResNet/Sequential[layer1]/Bottleneck[0]/Conv2d[conv2] %679 : Float(1, 64, 56, 56) = aten::batch_norm[training=0, momentum=0, eps=1e-05, cudnn_enabled=1](%674, %14, %15, %16, %17), scope: ResNet/Sequential[layer1]/Bottleneck[0]/BatchNorm2d[bn2] %681 : Float(1, 64, 56, 56) = aten::thresholdthreshold={0}, value={0}, scope: ResNet/Sequential[layer1]/Bottleneck[0]/ReLU[relu] %682 : Dynamic = prim::Undefined(), scope: ResNet/Sequential[layer1]/Bottleneck[0]/Conv2d[conv3] %691 : Float(1, 256, 56, 56) = aten::_convolution[stride=[1, 1], padding=[0, 0], dilation=[1, 1], transposed=0, output_padding=[0, 0], groups=1, benchmark=0, deterministic=0, cudnn_enabled=1](%681, %19, %682), scope: ResNet/Sequential[layer1]/Bottleneck[0]/Conv2d[conv3] %696 : Float(1, 256, 56, 56) = aten::batch_norm[training=0, momentum=0, eps=1e-05, cudnn_enabled=1](%691, %20, %21, %22, %23), scope: ResNet/Sequential[layer1]/Bottleneck[0]/BatchNorm2d[bn3] %697 : Dynamic = prim::Undefined(), scope: ResNet/Sequential[layer1]/Bottleneck[0]/Sequential[downsample]/Conv2d[0] %706 : Float(1, 256, 56, 56) = aten::_convolution[stride=[1, 1], padding=[0, 0], dilation=[1, 1], transposed=0, output_padding=[0, 0], groups=1, benchmark=0, deterministic=0, cudnn_enabled=1](%646, %25, %697), scope: ResNet/Sequential[layer1]/Bottleneck[0]/Sequential[downsample]/Conv2d[0] %711 : Float(1, 256, 56, 56) = aten::batch_norm[training=0, momentum=0, eps=1e-05, cudnn_enabled=1](%706, %26, %27, %28, %29), scope: ResNet/Sequential[layer1]/Bottleneck[0]/Sequential[downsample]/BatchNorm2d[1] %712 : Float(1, 256, 56, 56) = aten::add[alpha={1}](%696, %711), scope: ResNet/Sequential[layer1]/Bottleneck[0] %714 : Float(1, 256, 56, 56) = aten::thresholdthreshold={0}, value={0}, scope: ResNet/Sequential[layer1]/Bottleneck[0]/ReLU[relu] %715 : Dynamic = prim::Undefined(), scope: ResNet/Sequential[layer1]/Bottleneck[1]/Conv2d[conv1] %724 : Float(1, 64, 56, 56) = aten::_convolution[stride=[1, 1], padding=[0, 0], dilation=[1, 1], transposed=0, output_padding=[0, 0], groups=1, benchmark=0, deterministic=0, cudnn_enabled=1](%714, %31, %715), scope: ResNet/Sequential[layer1]/Bottleneck[1]/Conv2d[conv1] %729 : Float(1, 64, 56, 56) = aten::batch_norm[training=0, momentum=0, eps=1e-05, cudnn_enabled=1](%724, %32, %33, %34, %35), scope: ResNet/Sequential[layer1]/Bottleneck[1]/BatchNorm2d[bn1] %731 : Float(1, 64, 56, 56) = aten::thresholdthreshold={0}, value={0}, scope: ResNet/Sequential[layer1]/Bottleneck[1]/ReLU[relu] %732 : Dynamic = prim::Undefined(), scope: ResNet/Sequential[layer1]/Bottleneck[1]/Conv2d[conv2] %741 : Float(1, 64, 56, 56) = aten::_convolution[stride=[1, 1], padding=[1, 1], dilation=[1, 1], transposed=0, output_padding=[0, 0], groups=1, benchmark=0, deterministic=0, cudnn_enabled=1](%731, %37, %732), scope: ResNet/Sequential[layer1]/Bottleneck[1]/Conv2d[conv2] %746 : Float(1, 64, 56, 56) = aten::batch_norm[training=0, momentum=0, eps=1e-05, cudnn_enabled=1](%741, %38, %39, %40, %41), scope: ResNet/Sequential[layer1]/Bottleneck[1]/BatchNorm2d[bn2] %748 : Float(1, 64, 56, 56) = aten::thresholdthreshold={0}, value={0}, scope: ResNet/Sequential[layer1]/Bottleneck[1]/ReLU[relu] %749 : Dynamic = prim::Undefined(), scope: ResNet/Sequential[layer1]/Bottleneck[1]/Conv2d[conv3] %758 : Float(1, 256, 56, 56) = aten::_convolution[stride=[1, 1], padding=[0, 0], dilation=[1, 1], transposed=0, output_padding=[0, 0], groups=1, benchmark=0, deterministic=0, cudnn_enabled=1](%748, %43, %749), scope: ResNet/Sequential[layer1]/Bottleneck[1]/Conv2d[conv3] %763 : Float(1, 256, 56, 56) = aten::batch_norm[training=0, momentum=0, eps=1e-05, cudnn_enabled=1](%758, %44, %45, %46, %47), scope: ResNet/Sequential[layer1]/Bottleneck[1]/BatchNorm2d[bn3] %764 : Float(1, 256, 56, 56) = aten::add[alpha={1}](%763, %714), scope: ResNet/Sequential[layer1]/Bottleneck[1] %766 : Float(1, 256, 56, 56) = aten::thresholdthreshold={0}, value={0}, scope: ResNet/Sequential[layer1]/Bottleneck[1]/ReLU[relu] %767 : Dynamic = prim::Undefined(), scope: ResNet/Sequential[layer1]/Bottleneck[2]/Conv2d[conv1] %776 : Float(1, 64, 56, 56) = aten::_convolution[stride=[1, 1], padding=[0, 0], dilation=[1, 1], transposed=0, output_padding=[0, 0], groups=1, benchmark=0, deterministic=0, cudnn_enabled=1](%766, %49, %767), scope: ResNet/Sequential[layer1]/Bottleneck[2]/Conv2d[conv1] %781 : Float(1, 64, 56, 56) = aten::batch_norm[training=0, momentum=0, eps=1e-05, cudnn_enabled=1](%776, %50, %51, %52, %53), scope: ResNet/Sequential[layer1]/Bottleneck[2]/BatchNorm2d[bn1] %783 : Float(1, 64, 56, 56) = aten::thresholdthreshold={0}, value={0}, scope: ResNet/Sequential[layer1]/Bottleneck[2]/ReLU[relu] %784 : Dynamic = prim::Undefined(), scope: ResNet/Sequential[layer1]/Bottleneck[2]/Conv2d[conv2] %793 : Float(1, 64, 56, 56) = aten::_convolution[stride=[1, 1], padding=[1, 1], dilation=[1, 1], transposed=0, output_padding=[0, 0], groups=1, benchmark=0, deterministic=0, cudnn_enabled=1](%783, %55, %784), scope: ResNet/Sequential[layer1]/Bottleneck[2]/Conv2d[conv2] %798 : Float(1, 64, 56, 56) = aten::batch_norm[training=0, momentum=0, eps=1e-05, cudnn_enabled=1](%793, %56, %57, %58, %59), scope: ResNet/Sequential[layer1]/Bottleneck[2]/BatchNorm2d[bn2] %800 : Float(1, 64, 56, 56) = aten::thresholdthreshold={0}, value={0}, scope: ResNet/Sequential[layer1]/Bottleneck[2]/ReLU[relu] %801 : Dynamic = prim::Undefined(), scope: ResNet/Sequential[layer1]/Bottleneck[2]/Conv2d[conv3] %810 : Float(1, 256, 56, 56) = aten::_convolution[stride=[1, 1], padding=[0, 0], dilation=[1, 1], transposed=0, output_padding=[0, 0], groups=1, benchmark=0, deterministic=0, cudnn_enabled=1](%800, %61, %801), scope: ResNet/Sequential[layer1]/Bottleneck[2]/Conv2d[conv3] %815 : Float(1, 256, 56, 56) = aten::batch_norm[training=0, momentum=0, eps=1e-05, cudnn_enabled=1](%810, %62, %63, %64, %65), scope: ResNet/Sequential[layer1]/Bottleneck[2]/BatchNorm2d[bn3] %816 : Float(1, 256, 56, 56) = aten::add[alpha={1}](%815, %766), scope: ResNet/Sequential[layer1]/Bottleneck[2] %818 : Float(1, 256, 56, 56) = aten::thresholdthreshold={0}, value={0}, scope: ResNet/Sequential[layer1]/Bottleneck[2]/ReLU[relu] %819 : Dynamic = prim::Undefined(), scope: ResNet/Sequential[layer2]/Bottleneck[0]/Conv2d[conv1] %828 : Float(1, 128, 56, 56) = aten::_convolution[stride=[1, 1], padding=[0, 0], dilation=[1, 1], transposed=0, output_padding=[0, 0], groups=1, benchmark=0, deterministic=0, cudnn_enabled=1](%818, %67, %819), scope: ResNet/Sequential[layer2]/Bottleneck[0]/Conv2d[conv1] %833 : Float(1, 128, 56, 56) = aten::batch_norm[training=0, momentum=0, eps=1e-05, cudnn_enabled=1](%828, %68, %69, %70, %71), scope: ResNet/Sequential[layer2]/Bottleneck[0]/BatchNorm2d[bn1] %835 : Float(1, 128, 56, 56) = aten::thresholdthreshold={0}, value={0}, scope: ResNet/Sequential[layer2]/Bottleneck[0]/ReLU[relu] %836 : Dynamic = prim::Undefined(), scope: ResNet/Sequential[layer2]/Bottleneck[0]/Conv2d[conv2] %845 : Float(1, 128, 28, 28) = aten::_convolution[stride=[2, 2], padding=[1, 1], dilation=[1, 1], transposed=0, output_padding=[0, 0], groups=1, benchmark=0, deterministic=0, cudnn_enabled=1](%835, %73, %836), scope: ResNet/Sequential[layer2]/Bottleneck[0]/Conv2d[conv2] %850 : Float(1, 128, 28, 28) = aten::batch_norm[training=0, momentum=0, eps=1e-05, cudnn_enabled=1](%845, %74, %75, %76, %77), scope: ResNet/Sequential[layer2]/Bottleneck[0]/BatchNorm2d[bn2] %852 : Float(1, 128, 28, 28) = aten::thresholdthreshold={0}, value={0}, scope: ResNet/Sequential[layer2]/Bottleneck[0]/ReLU[relu] %853 : Dynamic = prim::Undefined(), scope: ResNet/Sequential[layer2]/Bottleneck[0]/Conv2d[conv3] %862 : Float(1, 512, 28, 28) = aten::_convolution[stride=[1, 1], padding=[0, 0], dilation=[1, 1], transposed=0, output_padding=[0, 0], groups=1, benchmark=0, deterministic=0, cudnn_enabled=1](%852, %79, %853), scope: ResNet/Sequential[layer2]/Bottleneck[0]/Conv2d[conv3] %867 : Float(1, 512, 28, 28) = aten::batch_norm[training=0, momentum=0, eps=1e-05, cudnn_enabled=1](%862, %80, %81, %82, %83), scope: ResNet/Sequential[layer2]/Bottleneck[0]/BatchNorm2d[bn3] %868 : Dynamic = prim::Undefined(), scope: ResNet/Sequential[layer2]/Bottleneck[0]/Sequential[downsample]/Conv2d[0] %877 : Float(1, 512, 28, 28) = aten::_convolution[stride=[2, 2], padding=[0, 0], dilation=[1, 1], transposed=0, output_padding=[0, 0], groups=1, benchmark=0, deterministic=0, cudnn_enabled=1](%818, %85, %868), scope: ResNet/Sequential[layer2]/Bottleneck[0]/Sequential[downsample]/Conv2d[0] %882 : Float(1, 512, 28, 28) = aten::batch_norm[training=0, momentum=0, eps=1e-05, cudnn_enabled=1](%877, %86, %87, %88, %89), scope: ResNet/Sequential[layer2]/Bottleneck[0]/Sequential[downsample]/BatchNorm2d[1] %883 : Float(1, 512, 28, 28) = aten::add[alpha={1}](%867, %882), scope: ResNet/Sequential[layer2]/Bottleneck[0] %885 : Float(1, 512, 28, 28) = aten::thresholdthreshold={0}, value={0}, scope: ResNet/Sequential[layer2]/Bottleneck[0]/ReLU[relu] %886 : Dynamic = prim::Undefined(), scope: ResNet/Sequential[layer2]/Bottleneck[1]/Conv2d[conv1] %895 : Float(1, 128, 28, 28) = aten::_convolution[stride=[1, 1], padding=[0, 0], dilation=[1, 1], transposed=0, output_padding=[0, 0], groups=1, benchmark=0, deterministic=0, cudnn_enabled=1](%885, %91, %886), scope: ResNet/Sequential[layer2]/Bottleneck[1]/Conv2d[conv1] %900 : Float(1, 128, 28, 28) = aten::batch_norm[training=0, momentum=0, eps=1e-05, cudnn_enabled=1](%895, %92, %93, %94, %95), scope: ResNet/Sequential[layer2]/Bottleneck[1]/BatchNorm2d[bn1] %902 : Float(1, 128, 28, 28) = aten::thresholdthreshold={0}, value={0}, scope: ResNet/Sequential[layer2]/Bottleneck[1]/ReLU[relu] %903 : Dynamic = prim::Undefined(), scope: ResNet/Sequential[layer2]/Bottleneck[1]/Conv2d[conv2] %912 : Float(1, 128, 28, 28) = aten::_convolution[stride=[1, 1], padding=[1, 1], dilation=[1, 1], transposed=0, output_padding=[0, 0], groups=1, benchmark=0, deterministic=0, cudnn_enabled=1](%902, %97, %903), scope: ResNet/Sequential[layer2]/Bottleneck[1]/Conv2d[conv2] %917 : Float(1, 128, 28, 28) = aten::batch_norm[training=0, momentum=0, eps=1e-05, cudnn_enabled=1](%912, %98, %99, %100, %101), scope: ResNet/Sequential[layer2]/Bottleneck[1]/BatchNorm2d[bn2] %919 : Float(1, 128, 28, 28) = aten::thresholdthreshold={0}, value={0}, scope: ResNet/Sequential[layer2]/Bottleneck[1]/ReLU[relu] %920 : Dynamic = prim::Undefined(), scope: ResNet/Sequential[layer2]/Bottleneck[1]/Conv2d[conv3] %929 : Float(1, 512, 28, 28) = aten::_convolution[stride=[1, 1], padding=[0, 0], dilation=[1, 1], transposed=0, output_padding=[0, 0], groups=1, benchmark=0, deterministic=0, cudnn_enabled=1](%919, %103, %920), scope: ResNet/Sequential[layer2]/Bottleneck[1]/Conv2d[conv3] %934 : Float(1, 512, 28, 28) = aten::batch_norm[training=0, momentum=0, eps=1e-05, cudnn_enabled=1](%929, %104, %105, %106, %107), scope: ResNet/Sequential[layer2]/Bottleneck[1]/BatchNorm2d[bn3] %935 : Float(1, 512, 28, 28) = aten::add[alpha={1}](%934, %885), scope: ResNet/Sequential[layer2]/Bottleneck[1] %937 : Float(1, 512, 28, 28) = aten::thresholdthreshold={0}, value={0}, scope: ResNet/Sequential[layer2]/Bottleneck[1]/ReLU[relu] %938 : Dynamic = prim::Undefined(), scope: ResNet/Sequential[layer2]/Bottleneck[2]/Conv2d[conv1] %947 : Float(1, 128, 28, 28) = aten::_convolution[stride=[1, 1], padding=[0, 0], dilation=[1, 1], transposed=0, output_padding=[0, 0], groups=1, benchmark=0, deterministic=0, cudnn_enabled=1](%937, %109, %938), scope: ResNet/Sequential[layer2]/Bottleneck[2]/Conv2d[conv1] %952 : Float(1, 128, 28, 28) = aten::batch_norm[training=0, momentum=0, eps=1e-05, cudnn_enabled=1](%947, %110, %111, %112, %113), scope: ResNet/Sequential[layer2]/Bottleneck[2]/BatchNorm2d[bn1] %954 : Float(1, 128, 28, 28) = aten::thresholdthreshold={0}, value={0}, scope: ResNet/Sequential[layer2]/Bottleneck[2]/ReLU[relu] %955 : Dynamic = prim::Undefined(), scope: ResNet/Sequential[layer2]/Bottleneck[2]/Conv2d[conv2] %964 : Float(1, 128, 28, 28) = aten::_convolution[stride=[1, 1], padding=[1, 1], dilation=[1, 1], transposed=0, output_padding=[0, 0], groups=1, benchmark=0, deterministic=0, cudnn_enabled=1](%954, %115, %955), scope: ResNet/Sequential[layer2]/Bottleneck[2]/Conv2d[conv2] %969 : Float(1, 128, 28, 28) = aten::batch_norm[training=0, momentum=0, eps=1e-05, cudnn_enabled=1](%964, %116, %117, %118, %119), scope: ResNet/Sequential[layer2]/Bottleneck[2]/BatchNorm2d[bn2] %971 : Float(1, 128, 28, 28) = aten::thresholdthreshold={0}, value={0}, scope: ResNet/Sequential[layer2]/Bottleneck[2]/ReLU[relu] %972 : Dynamic = prim::Undefined(), scope: ResNet/Sequential[layer2]/Bottleneck[2]/Conv2d[conv3] %981 : Float(1, 512, 28, 28) = aten::_convolution[stride=[1, 1], padding=[0, 0], dilation=[1, 1], transposed=0, output_padding=[0, 0], groups=1, benchmark=0, deterministic=0, cudnn_enabled=1](%971, %121, %972), scope: ResNet/Sequential[layer2]/Bottleneck[2]/Conv2d[conv3] %986 : Float(1, 512, 28, 28) = aten::batch_norm[training=0, momentum=0, eps=1e-05, cudnn_enabled=1](%981, %122, %123, %124, %125), scope: ResNet/Sequential[layer2]/Bottleneck[2]/BatchNorm2d[bn3] %987 : Float(1, 512, 28, 28) = aten::add[alpha={1}](%986, %937), scope: ResNet/Sequential[layer2]/Bottleneck[2] %989 : Float(1, 512, 28, 28) = aten::thresholdthreshold={0}, value={0}, scope: ResNet/Sequential[layer2]/Bottleneck[2]/ReLU[relu] %990 : Dynamic = prim::Undefined(), scope: ResNet/Sequential[layer2]/Bottleneck[3]/Conv2d[conv1] %999 : Float(1, 128, 28, 28) = aten::_convolution[stride=[1, 1], padding=[0, 0], dilation=[1, 1], transposed=0, output_padding=[0, 0], groups=1, benchmark=0, deterministic=0, cudnn_enabled=1](%989, %127, %990), scope: ResNet/Sequential[layer2]/Bottleneck[3]/Conv2d[conv1] %1004 : Float(1, 128, 28, 28) = aten::batch_norm[training=0, momentum=0, eps=1e-05, cudnn_enabled=1](%999, %128, %129, %130, %131), scope: ResNet/Sequential[layer2]/Bottleneck[3]/BatchNorm2d[bn1] %1006 : Float(1, 128, 28, 28) = aten::thresholdthreshold={0}, value={0}, scope: ResNet/Sequential[layer2]/Bottleneck[3]/ReLU[relu] %1007 : Dynamic = prim::Undefined(), scope: ResNet/Sequential[layer2]/Bottleneck[3]/Conv2d[conv2] %1016 : Float(1, 128, 28, 28) = aten::_convolution[stride=[1, 1], padding=[1, 1], dilation=[1, 1], transposed=0, output_padding=[0, 0], groups=1, benchmark=0, deterministic=0, cudnn_enabled=1](%1006, %133, %1007), scope: ResNet/Sequential[layer2]/Bottleneck[3]/Conv2d[conv2] %1021 : Float(1, 128, 28, 28) = aten::batch_norm[training=0, momentum=0, eps=1e-05, cudnn_enabled=1](%1016, %134, %135, %136, %137), scope: ResNet/Sequential[layer2]/Bottleneck[3]/BatchNorm2d[bn2] %1023 : Float(1, 128, 28, 28) = aten::thresholdthreshold={0}, value={0}, scope: ResNet/Sequential[layer2]/Bottleneck[3]/ReLU[relu] %1024 : Dynamic = prim::Undefined(), scope: ResNet/Sequential[layer2]/Bottleneck[3]/Conv2d[conv3] %1033 : Float(1, 512, 28, 28) = aten::_convolution[stride=[1, 1], padding=[0, 0], dilation=[1, 1], transposed=0, output_padding=[0, 0], groups=1, benchmark=0, deterministic=0, cudnn_enabled=1](%1023, %139, %1024), scope: ResNet/Sequential[layer2]/Bottleneck[3]/Conv2d[conv3] %1038 : Float(1, 512, 28, 28) = aten::batch_norm[training=0, momentum=0, eps=1e-05, cudnn_enabled=1](%1033, %140, %141, %142, %143), scope: ResNet/Sequential[layer2]/Bottleneck[3]/BatchNorm2d[bn3] %1039 : Float(1, 512, 28, 28) = aten::add[alpha={1}](%1038, %989), scope: ResNet/Sequential[layer2]/Bottleneck[3] %1041 : Float(1, 512, 28, 28) = aten::thresholdthreshold={0}, value={0}, scope: ResNet/Sequential[layer2]/Bottleneck[3]/ReLU[relu] %1042 : Dynamic = prim::Undefined(), scope: ResNet/Sequential[layer3]/Bottleneck[0]/Conv2d[conv1] %1051 : Float(1, 256, 28, 28) = aten::_convolution[stride=[1, 1], padding=[0, 0], dilation=[1, 1], transposed=0, output_padding=[0, 0], groups=1, benchmark=0, deterministic=0, cudnn_enabled=1](%1041, %145, %1042), scope: ResNet/Sequential[layer3]/Bottleneck[0]/Conv2d[conv1] %1056 : Float(1, 256, 28, 28) = aten::batch_norm[training=0, momentum=0, eps=1e-05, cudnn_enabled=1](%1051, %146, %147, %148, %149), scope: ResNet/Sequential[layer3]/Bottleneck[0]/BatchNorm2d[bn1] %1058 : Float(1, 256, 28, 28) = aten::thresholdthreshold={0}, value={0}, scope: ResNet/Sequential[layer3]/Bottleneck[0]/ReLU[relu] %1059 : Dynamic = prim::Undefined(), scope: ResNet/Sequential[layer3]/Bottleneck[0]/Conv2d[conv2] %1068 : Float(1, 256, 14, 14) = aten::_convolution[stride=[2, 2], padding=[1, 1], dilation=[1, 1], transposed=0, output_padding=[0, 0], groups=1, benchmark=0, deterministic=0, cudnn_enabled=1](%1058, %151, %1059), scope: ResNet/Sequential[layer3]/Bottleneck[0]/Conv2d[conv2] %1073 : Float(1, 256, 14, 14) = aten::batch_norm[training=0, momentum=0, eps=1e-05, cudnn_enabled=1](%1068, %152, %153, %154, %155), scope: ResNet/Sequential[layer3]/Bottleneck[0]/BatchNorm2d[bn2] %1075 : Float(1, 256, 14, 14) = aten::thresholdthreshold={0}, value={0}, scope: ResNet/Sequential[layer3]/Bottleneck[0]/ReLU[relu] %1076 : Dynamic = prim::Undefined(), scope: ResNet/Sequential[layer3]/Bottleneck[0]/Conv2d[conv3] %1085 : Float(1, 1024, 14, 14) = aten::_convolution[stride=[1, 1], padding=[0, 0], dilation=[1, 1], transposed=0, output_padding=[0, 0], groups=1, benchmark=0, deterministic=0, cudnn_enabled=1](%1075, %157, %1076), scope: ResNet/Sequential[layer3]/Bottleneck[0]/Conv2d[conv3] %1090 : Float(1, 1024, 14, 14) = aten::batch_norm[training=0, momentum=0, eps=1e-05, cudnn_enabled=1](%1085, %158, %159, %160, %161), scope: ResNet/Sequential[layer3]/Bottleneck[0]/BatchNorm2d[bn3] %1091 : Dynamic = prim::Undefined(), scope: ResNet/Sequential[layer3]/Bottleneck[0]/Sequential[downsample]/Conv2d[0] %1100 : Float(1, 1024, 14, 14) = aten::_convolution[stride=[2, 2], padding=[0, 0], dilation=[1, 1], transposed=0, output_padding=[0, 0], groups=1, benchmark=0, deterministic=0, cudnn_enabled=1](%1041, %163, %1091), scope: ResNet/Sequential[layer3]/Bottleneck[0]/Sequential[downsample]/Conv2d[0] %1105 : Float(1, 1024, 14, 14) = aten::batch_norm[training=0, momentum=0, eps=1e-05, cudnn_enabled=1](%1100, %164, %165, %166, %167), scope: ResNet/Sequential[layer3]/Bottleneck[0]/Sequential[downsample]/BatchNorm2d[1] %1106 : Float(1, 1024, 14, 14) = aten::add[alpha={1}](%1090, %1105), scope: ResNet/Sequential[layer3]/Bottleneck[0] %1108 : Float(1, 1024, 14, 14) = aten::thresholdthreshold={0}, value={0}, scope: ResNet/Sequential[layer3]/Bottleneck[0]/ReLU[relu] %1109 : Dynamic = prim::Undefined(), scope: ResNet/Sequential[layer3]/Bottleneck[1]/Conv2d[conv1] %1118 : Float(1, 256, 14, 14) = aten::_convolution[stride=[1, 1], padding=[0, 0], dilation=[1, 1], transposed=0, output_padding=[0, 0], groups=1, benchmark=0, deterministic=0, cudnn_enabled=1](%1108, %169, %1109), scope: ResNet/Sequential[layer3]/Bottleneck[1]/Conv2d[conv1] %1123 : Float(1, 256, 14, 14) = aten::batch_norm[training=0, momentum=0, eps=1e-05, cudnn_enabled=1](%1118, %170, %171, %172, %173), scope: ResNet/Sequential[layer3]/Bottleneck[1]/BatchNorm2d[bn1] %1125 : Float(1, 256, 14, 14) = aten::thresholdthreshold={0}, value={0}, scope: ResNet/Sequential[layer3]/Bottleneck[1]/ReLU[relu] %1126 : Dynamic = prim::Undefined(), scope: ResNet/Sequential[layer3]/Bottleneck[1]/Conv2d[conv2] %1135 : Float(1, 256, 14, 14) = aten::_convolution[stride=[1, 1], padding=[1, 1], dilation=[1, 1], transposed=0, output_padding=[0, 0], groups=1, benchmark=0, deterministic=0, cudnn_enabled=1](%1125, %175, %1126), scope: ResNet/Sequential[layer3]/Bottleneck[1]/Conv2d[conv2] %1140 : Float(1, 256, 14, 14) = aten::batch_norm[training=0, momentum=0, eps=1e-05, cudnn_enabled=1](%1135, %176, %177, %178, %179), scope: ResNet/Sequential[layer3]/Bottleneck[1]/BatchNorm2d[bn2] %1142 : Float(1, 256, 14, 14) = aten::thresholdthreshold={0}, value={0}, scope: ResNet/Sequential[layer3]/Bottleneck[1]/ReLU[relu] %1143 : Dynamic = prim::Undefined(), scope: ResNet/Sequential[layer3]/Bottleneck[1]/Conv2d[conv3] %1152 : Float(1, 1024, 14, 14) = aten::_convolution[stride=[1, 1], padding=[0, 0], dilation=[1, 1], transposed=0, output_padding=[0, 0], groups=1, benchmark=0, deterministic=0, cudnn_enabled=1](%1142, %181, %1143), scope: ResNet/Sequential[layer3]/Bottleneck[1]/Conv2d[conv3] %1157 : Float(1, 1024, 14, 14) = aten::batch_norm[training=0, momentum=0, eps=1e-05, cudnn_enabled=1](%1152, %182, %183, %184, %185), scope: ResNet/Sequential[layer3]/Bottleneck[1]/BatchNorm2d[bn3] %1158 : Float(1, 1024, 14, 14) = aten::add[alpha={1}](%1157, %1108), scope: ResNet/Sequential[layer3]/Bottleneck[1] %1160 : Float(1, 1024, 14, 14) = aten::thresholdthreshold={0}, value={0}, scope: ResNet/Sequential[layer3]/Bottleneck[1]/ReLU[relu] %1161 : Dynamic = prim::Undefined(), scope: ResNet/Sequential[layer3]/Bottleneck[2]/Conv2d[conv1] %1170 : Float(1, 256, 14, 14) = aten::_convolution[stride=[1, 1], padding=[0, 0], dilation=[1, 1], transposed=0, output_padding=[0, 0], groups=1, benchmark=0, deterministic=0, cudnn_enabled=1](%1160, %187, %1161), scope: ResNet/Sequential[layer3]/Bottleneck[2]/Conv2d[conv1] %1175 : Float(1, 256, 14, 14) = aten::batch_norm[training=0, momentum=0, eps=1e-05, cudnn_enabled=1](%1170, %188, %189, %190, %191), scope: ResNet/Sequential[layer3]/Bottleneck[2]/BatchNorm2d[bn1] %1177 : Float(1, 256, 14, 14) = aten::thresholdthreshold={0}, value={0}, scope: ResNet/Sequential[layer3]/Bottleneck[2]/ReLU[relu] %1178 : Dynamic = prim::Undefined(), scope: ResNet/Sequential[layer3]/Bottleneck[2]/Conv2d[conv2] %1187 : Float(1, 256, 14, 14) = aten::_convolution[stride=[1, 1], padding=[1, 1], dilation=[1, 1], transposed=0, output_padding=[0, 0], groups=1, benchmark=0, deterministic=0, cudnn_enabled=1](%1177, %193, %1178), scope: ResNet/Sequential[layer3]/Bottleneck[2]/Conv2d[conv2] %1192 : Float(1, 256, 14, 14) = aten::batch_norm[training=0, momentum=0, eps=1e-05, cudnn_enabled=1](%1187, %194, %195, %196, %197), scope: ResNet/Sequential[layer3]/Bottleneck[2]/BatchNorm2d[bn2] %1194 : Float(1, 256, 14, 14) = aten::thresholdthreshold={0}, value={0}, scope: ResNet/Sequential[layer3]/Bottleneck[2]/ReLU[relu] %1195 : Dynamic = prim::Undefined(), scope: ResNet/Sequential[layer3]/Bottleneck[2]/Conv2d[conv3] %1204 : Float(1, 1024, 14, 14) = aten::_convolution[stride=[1, 1], padding=[0, 0], dilation=[1, 1], transposed=0, output_padding=[0, 0], groups=1, benchmark=0, deterministic=0, cudnn_enabled=1](%1194, %199, %1195), scope: ResNet/Sequential[layer3]/Bottleneck[2]/Conv2d[conv3] %1209 : Float(1, 1024, 14, 14) = aten::batch_norm[training=0, momentum=0, eps=1e-05, cudnn_enabled=1](%1204, %200, %201, %202, %203), scope: ResNet/Sequential[layer3]/Bottleneck[2]/BatchNorm2d[bn3] %1210 : Float(1, 1024, 14, 14) = aten::add[alpha={1}](%1209, %1160), scope: ResNet/Sequential[layer3]/Bottleneck[2] %1212 : Float(1, 1024, 14, 14) = aten::thresholdthreshold={0}, value={0}, scope: ResNet/Sequential[layer3]/Bottleneck[2]/ReLU[relu] %1213 : Dynamic = prim::Undefined(), scope: ResNet/Sequential[layer3]/Bottleneck[3]/Conv2d[conv1] %1222 : Float(1, 256, 14, 14) = aten::_convolution[stride=[1, 1], padding=[0, 0], dilation=[1, 1], transposed=0, output_padding=[0, 0], groups=1, benchmark=0, deterministic=0, cudnn_enabled=1](%1212, %205, %1213), scope: ResNet/Sequential[layer3]/Bottleneck[3]/Conv2d[conv1] %1227 : Float(1, 256, 14, 14) = aten::batch_norm[training=0, momentum=0, eps=1e-05, cudnn_enabled=1](%1222, %206, %207, %208, %209), scope: ResNet/Sequential[layer3]/Bottleneck[3]/BatchNorm2d[bn1] %1229 : Float(1, 256, 14, 14) = aten::thresholdthreshold={0}, value={0}, scope: ResNet/Sequential[layer3]/Bottleneck[3]/ReLU[relu] %1230 : Dynamic = prim::Undefined(), scope: ResNet/Sequential[layer3]/Bottleneck[3]/Conv2d[conv2] %1239 : Float(1, 256, 14, 14) = aten::_convolution[stride=[1, 1], padding=[1, 1], dilation=[1, 1], transposed=0, output_padding=[0, 0], groups=1, benchmark=0, deterministic=0, cudnn_enabled=1](%1229, %211, %1230), scope: ResNet/Sequential[layer3]/Bottleneck[3]/Conv2d[conv2] %1244 : Float(1, 256, 14, 14) = aten::batch_norm[training=0, momentum=0, eps=1e-05, cudnn_enabled=1](%1239, %212, %213, %214, %215), scope: ResNet/Sequential[layer3]/Bottleneck[3]/BatchNorm2d[bn2] %1246 : Float(1, 256, 14, 14) = aten::thresholdthreshold={0}, value={0}, scope: ResNet/Sequential[layer3]/Bottleneck[3]/ReLU[relu] %1247 : Dynamic = prim::Undefined(), scope: ResNet/Sequential[layer3]/Bottleneck[3]/Conv2d[conv3] %1256 : Float(1, 1024, 14, 14) = aten::_convolution[stride=[1, 1], padding=[0, 0], dilation=[1, 1], transposed=0, output_padding=[0, 0], groups=1, benchmark=0, deterministic=0, cudnn_enabled=1](%1246, %217, %1247), scope: ResNet/Sequential[layer3]/Bottleneck[3]/Conv2d[conv3] %1261 : Float(1, 1024, 14, 14) = aten::batch_norm[training=0, momentum=0, eps=1e-05, cudnn_enabled=1](%1256, %218, %219, %220, %221), scope: ResNet/Sequential[layer3]/Bottleneck[3]/BatchNorm2d[bn3] %1262 : Float(1, 1024, 14, 14) = aten::add[alpha={1}](%1261, %1212), scope: ResNet/Sequential[layer3]/Bottleneck[3] %1264 : Float(1, 1024, 14, 14) = aten::thresholdthreshold={0}, value={0}, scope: ResNet/Sequential[layer3]/Bottleneck[3]/ReLU[relu] %1265 : Dynamic = prim::Undefined(), scope: ResNet/Sequential[layer3]/Bottleneck[4]/Conv2d[conv1] %1274 : Float(1, 256, 14, 14) = aten::_convolution[stride=[1, 1], padding=[0, 0], dilation=[1, 1], transposed=0, output_padding=[0, 0], groups=1, benchmark=0, deterministic=0, cudnn_enabled=1](%1264, %223, %1265), scope: ResNet/Sequential[layer3]/Bottleneck[4]/Conv2d[conv1] %1279 : Float(1, 256, 14, 14) = aten::batch_norm[training=0, momentum=0, eps=1e-05, cudnn_enabled=1](%1274, %224, %225, %226, %227), scope: ResNet/Sequential[layer3]/Bottleneck[4]/BatchNorm2d[bn1] %1281 : Float(1, 256, 14, 14) = aten::thresholdthreshold={0}, value={0}, scope: ResNet/Sequential[layer3]/Bottleneck[4]/ReLU[relu] %1282 : Dynamic = prim::Undefined(), scope: ResNet/Sequential[layer3]/Bottleneck[4]/Conv2d[conv2] %1291 : Float(1, 256, 14, 14) = aten::_convolution[stride=[1, 1], padding=[1, 1], dilation=[1, 1], transposed=0, output_padding=[0, 0], groups=1, benchmark=0, deterministic=0, cudnn_enabled=1](%1281, %229, %1282), scope: ResNet/Sequential[layer3]/Bottleneck[4]/Conv2d[conv2] %1296 : Float(1, 256, 14, 14) = aten::batch_norm[training=0, momentum=0, eps=1e-05, cudnn_enabled=1](%1291, %230, %231, %232, %233), scope: ResNet/Sequential[layer3]/Bottleneck[4]/BatchNorm2d[bn2] %1298 : Float(1, 256, 14, 14) = aten::thresholdthreshold={0}, value={0}, scope: ResNet/Sequential[layer3]/Bottleneck[4]/ReLU[relu] %1299 : Dynamic = prim::Undefined(), scope: ResNet/Sequential[layer3]/Bottleneck[4]/Conv2d[conv3] %1308 : Float(1, 1024, 14, 14) = aten::_convolution[stride=[1, 1], padding=[0, 0], dilation=[1, 1], transposed=0, output_padding=[0, 0], groups=1, benchmark=0, deterministic=0, cudnn_enabled=1](%1298, %235, %1299), scope: ResNet/Sequential[layer3]/Bottleneck[4]/Conv2d[conv3] %1313 : Float(1, 1024, 14, 14) = aten::batch_norm[training=0, momentum=0, eps=1e-05, cudnn_enabled=1](%1308, %236, %237, %238, %239), scope: ResNet/Sequential[layer3]/Bottleneck[4]/BatchNorm2d[bn3] %1314 : Float(1, 1024, 14, 14) = aten::add[alpha={1}](%1313, %1264), scope: ResNet/Sequential[layer3]/Bottleneck[4] %1316 : Float(1, 1024, 14, 14) = aten::thresholdthreshold={0}, value={0}, scope: ResNet/Sequential[layer3]/Bottleneck[4]/ReLU[relu] %1317 : Dynamic = prim::Undefined(), scope: ResNet/Sequential[layer3]/Bottleneck[5]/Conv2d[conv1] %1326 : Float(1, 256, 14, 14) = aten::_convolution[stride=[1, 1], padding=[0, 0], dilation=[1, 1], transposed=0, output_padding=[0, 0], groups=1, benchmark=0, deterministic=0, cudnn_enabled=1](%1316, %241, %1317), scope: ResNet/Sequential[layer3]/Bottleneck[5]/Conv2d[conv1] %1331 : Float(1, 256, 14, 14) = aten::batch_norm[training=0, momentum=0, eps=1e-05, cudnn_enabled=1](%1326, %242, %243, %244, %245), scope: ResNet/Sequential[layer3]/Bottleneck[5]/BatchNorm2d[bn1] %1333 : Float(1, 256, 14, 14) = aten::thresholdthreshold={0}, value={0}, scope: ResNet/Sequential[layer3]/Bottleneck[5]/ReLU[relu] %1334 : Dynamic = prim::Undefined(), scope: ResNet/Sequential[layer3]/Bottleneck[5]/Conv2d[conv2] %1343 : Float(1, 256, 14, 14) = aten::_convolution[stride=[1, 1], padding=[1, 1], dilation=[1, 1], transposed=0, output_padding=[0, 0], groups=1, benchmark=0, deterministic=0, cudnn_enabled=1](%1333, %247, %1334), scope: ResNet/Sequential[layer3]/Bottleneck[5]/Conv2d[conv2] %1348 : Float(1, 256, 14, 14) = aten::batch_norm[training=0, momentum=0, eps=1e-05, cudnn_enabled=1](%1343, %248, %249, %250, %251), scope: ResNet/Sequential[layer3]/Bottleneck[5]/BatchNorm2d[bn2] %1350 : Float(1, 256, 14, 14) = aten::thresholdthreshold={0}, value={0}, scope: ResNet/Sequential[layer3]/Bottleneck[5]/ReLU[relu] %1351 : Dynamic = prim::Undefined(), scope: ResNet/Sequential[layer3]/Bottleneck[5]/Conv2d[conv3] %1360 : Float(1, 1024, 14, 14) = aten::_convolution[stride=[1, 1], padding=[0, 0], dilation=[1, 1], transposed=0, output_padding=[0, 0], groups=1, benchmark=0, deterministic=0, cudnn_enabled=1](%1350, %253, %1351), scope: ResNet/Sequential[layer3]/Bottleneck[5]/Conv2d[conv3] %1365 : Float(1, 1024, 14, 14) = aten::batch_norm[training=0, momentum=0, eps=1e-05, cudnn_enabled=1](%1360, %254, %255, %256, %257), scope: ResNet/Sequential[layer3]/Bottleneck[5]/BatchNorm2d[bn3] %1366 : Float(1, 1024, 14, 14) = aten::add[alpha={1}](%1365, %1316), scope: ResNet/Sequential[layer3]/Bottleneck[5] %1368 : Float(1, 1024, 14, 14) = aten::thresholdthreshold={0}, value={0}, scope: ResNet/Sequential[layer3]/Bottleneck[5]/ReLU[relu] %1369 : Dynamic = prim::Undefined(), scope: ResNet/Sequential[layer3]/Bottleneck[6]/Conv2d[conv1] %1378 : Float(1, 256, 14, 14) = aten::_convolution[stride=[1, 1], padding=[0, 0], dilation=[1, 1], transposed=0, output_padding=[0, 0], groups=1, benchmark=0, deterministic=0, cudnn_enabled=1](%1368, %259, %1369), scope: ResNet/Sequential[layer3]/Bottleneck[6]/Conv2d[conv1] %1383 : Float(1, 256, 14, 14) = aten::batch_norm[training=0, momentum=0, eps=1e-05, cudnn_enabled=1](%1378, %260, %261, %262, %263), scope: ResNet/Sequential[layer3]/Bottleneck[6]/BatchNorm2d[bn1] %1385 : Float(1, 256, 14, 14) = aten::thresholdthreshold={0}, value={0}, scope: ResNet/Sequential[layer3]/Bottleneck[6]/ReLU[relu] %1386 : Dynamic = prim::Undefined(), scope: ResNet/Sequential[layer3]/Bottleneck[6]/Conv2d[conv2] %1395 : Float(1, 256, 14, 14) = aten::_convolution[stride=[1, 1], padding=[1, 1], dilation=[1, 1], transposed=0, output_padding=[0, 0], groups=1, benchmark=0, deterministic=0, cudnn_enabled=1](%1385, %265, %1386), scope: ResNet/Sequential[layer3]/Bottleneck[6]/Conv2d[conv2] %1400 : Float(1, 256, 14, 14) = aten::batch_norm[training=0, momentum=0, eps=1e-05, cudnn_enabled=1](%1395, %266, %267, %268, %269), scope: ResNet/Sequential[layer3]/Bottleneck[6]/BatchNorm2d[bn2] %1402 : Float(1, 256, 14, 14) = aten::thresholdthreshold={0}, value={0}, scope: ResNet/Sequential[layer3]/Bottleneck[6]/ReLU[relu] %1403 : Dynamic = prim::Undefined(), scope: ResNet/Sequential[layer3]/Bottleneck[6]/Conv2d[conv3] %1412 : Float(1, 1024, 14, 14) = aten::_convolution[stride=[1, 1], padding=[0, 0], dilation=[1, 1], transposed=0, output_padding=[0, 0], groups=1, benchmark=0, deterministic=0, cudnn_enabled=1](%1402, %271, %1403), scope: ResNet/Sequential[layer3]/Bottleneck[6]/Conv2d[conv3] %1417 : Float(1, 1024, 14, 14) = aten::batch_norm[training=0, momentum=0, eps=1e-05, cudnn_enabled=1](%1412, %272, %273, %274, %275), scope: ResNet/Sequential[layer3]/Bottleneck[6]/BatchNorm2d[bn3] %1418 : Float(1, 1024, 14, 14) = aten::add[alpha={1}](%1417, %1368), scope: ResNet/Sequential[layer3]/Bottleneck[6] %1420 : Float(1, 1024, 14, 14) = aten::thresholdthreshold={0}, value={0}, scope: ResNet/Sequential[layer3]/Bottleneck[6]/ReLU[relu] %1421 : Dynamic = prim::Undefined(), scope: ResNet/Sequential[layer3]/Bottleneck[7]/Conv2d[conv1] %1430 : Float(1, 256, 14, 14) = aten::_convolution[stride=[1, 1], padding=[0, 0], dilation=[1, 1], transposed=0, output_padding=[0, 0], groups=1, benchmark=0, deterministic=0, cudnn_enabled=1](%1420, %277, %1421), scope: ResNet/Sequential[layer3]/Bottleneck[7]/Conv2d[conv1] %1435 : Float(1, 256, 14, 14) = aten::batch_norm[training=0, momentum=0, eps=1e-05, cudnn_enabled=1](%1430, %278, %279, %280, %281), scope: ResNet/Sequential[layer3]/Bottleneck[7]/BatchNorm2d[bn1] %1437 : Float(1, 256, 14, 14) = aten::thresholdthreshold={0}, value={0}, scope: ResNet/Sequential[layer3]/Bottleneck[7]/ReLU[relu] %1438 : Dynamic = prim::Undefined(), scope: ResNet/Sequential[layer3]/Bottleneck[7]/Conv2d[conv2] %1447 : Float(1, 256, 14, 14) = aten::_convolution[stride=[1, 1], padding=[1, 1], dilation=[1, 1], transposed=0, output_padding=[0, 0], groups=1, benchmark=0, deterministic=0, cudnn_enabled=1](%1437, %283, %1438), scope: ResNet/Sequential[layer3]/Bottleneck[7]/Conv2d[conv2] %1452 : Float(1, 256, 14, 14) = aten::batch_norm[training=0, momentum=0, eps=1e-05, cudnn_enabled=1](%1447, %284, %285, %286, %287), scope: ResNet/Sequential[layer3]/Bottleneck[7]/BatchNorm2d[bn2] %1454 : Float(1, 256, 14, 14) = aten::thresholdthreshold={0}, value={0}, scope: ResNet/Sequential[layer3]/Bottleneck[7]/ReLU[relu] %1455 : Dynamic = prim::Undefined(), scope: ResNet/Sequential[layer3]/Bottleneck[7]/Conv2d[conv3] %1464 : Float(1, 1024, 14, 14) = aten::_convolution[stride=[1, 1], padding=[0, 0], dilation=[1, 1], transposed=0, output_padding=[0, 0], groups=1, benchmark=0, deterministic=0, cudnn_enabled=1](%1454, %289, %1455), scope: ResNet/Sequential[layer3]/Bottleneck[7]/Conv2d[conv3] %1469 : Float(1, 1024, 14, 14) = aten::batch_norm[training=0, momentum=0, eps=1e-05, cudnn_enabled=1](%1464, %290, %291, %292, %293), scope: ResNet/Sequential[layer3]/Bottleneck[7]/BatchNorm2d[bn3] %1470 : Float(1, 1024, 14, 14) = aten::add[alpha={1}](%1469, %1420), scope: ResNet/Sequential[layer3]/Bottleneck[7] %1472 : Float(1, 1024, 14, 14) = aten::thresholdthreshold={0}, value={0}, scope: ResNet/Sequential[layer3]/Bottleneck[7]/ReLU[relu] %1473 : Dynamic = prim::Undefined(), scope: ResNet/Sequential[layer3]/Bottleneck[8]/Conv2d[conv1] %1482 : Float(1, 256, 14, 14) = aten::_convolution[stride=[1, 1], padding=[0, 0], dilation=[1, 1], transposed=0, output_padding=[0, 0], groups=1, benchmark=0, deterministic=0, cudnn_enabled=1](%1472, %295, %1473), scope: ResNet/Sequential[layer3]/Bottleneck[8]/Conv2d[conv1] %1487 : Float(1, 256, 14, 14) = aten::batch_norm[training=0, momentum=0, eps=1e-05, cudnn_enabled=1](%1482, %296, %297, %298, %299), scope: ResNet/Sequential[layer3]/Bottleneck[8]/BatchNorm2d[bn1] %1489 : Float(1, 256, 14, 14) = aten::thresholdthreshold={0}, value={0}, scope: ResNet/Sequential[layer3]/Bottleneck[8]/ReLU[relu] %1490 : Dynamic = prim::Undefined(), scope: ResNet/Sequential[layer3]/Bottleneck[8]/Conv2d[conv2] %1499 : Float(1, 256, 14, 14) = aten::_convolution[stride=[1, 1], padding=[1, 1], dilation=[1, 1], transposed=0, output_padding=[0, 0], groups=1, benchmark=0, deterministic=0, cudnn_enabled=1](%1489, %301, %1490), scope: ResNet/Sequential[layer3]/Bottleneck[8]/Conv2d[conv2] %1504 : Float(1, 256, 14, 14) = aten::batch_norm[training=0, momentum=0, eps=1e-05, cudnn_enabled=1](%1499, %302, %303, %304, %305), scope: ResNet/Sequential[layer3]/Bottleneck[8]/BatchNorm2d[bn2] %1506 : Float(1, 256, 14, 14) = aten::thresholdthreshold={0}, value={0}, scope: ResNet/Sequential[layer3]/Bottleneck[8]/ReLU[relu] %1507 : Dynamic = prim::Undefined(), scope: ResNet/Sequential[layer3]/Bottleneck[8]/Conv2d[conv3] %1516 : Float(1, 1024, 14, 14) = aten::_convolution[stride=[1, 1], padding=[0, 0], dilation=[1, 1], transposed=0, output_padding=[0, 0], groups=1, benchmark=0, deterministic=0, cudnn_enabled=1](%1506, %307, %1507), scope: ResNet/Sequential[layer3]/Bottleneck[8]/Conv2d[conv3] %1521 : Float(1, 1024, 14, 14) = aten::batch_norm[training=0, momentum=0, eps=1e-05, cudnn_enabled=1](%1516, %308, %309, %310, %311), scope: ResNet/Sequential[layer3]/Bottleneck[8]/BatchNorm2d[bn3] %1522 : Float(1, 1024, 14, 14) = aten::add[alpha={1}](%1521, %1472), scope: ResNet/Sequential[layer3]/Bottleneck[8] %1524 : Float(1, 1024, 14, 14) = aten::thresholdthreshold={0}, value={0}, scope: ResNet/Sequential[layer3]/Bottleneck[8]/ReLU[relu] %1525 : Dynamic = prim::Undefined(), scope: ResNet/Sequential[layer3]/Bottleneck[9]/Conv2d[conv1] %1534 : Float(1, 256, 14, 14) = aten::_convolution[stride=[1, 1], padding=[0, 0], dilation=[1, 1], transposed=0, output_padding=[0, 0], groups=1, benchmark=0, deterministic=0, cudnn_enabled=1](%1524, %313, %1525), scope: ResNet/Sequential[layer3]/Bottleneck[9]/Conv2d[conv1] %1539 : Float(1, 256, 14, 14) = aten::batch_norm[training=0, momentum=0, eps=1e-05, cudnn_enabled=1](%1534, %314, %315, %316, %317), scope: ResNet/Sequential[layer3]/Bottleneck[9]/BatchNorm2d[bn1] %1541 : Float(1, 256, 14, 14) = aten::thresholdthreshold={0}, value={0}, scope: ResNet/Sequential[layer3]/Bottleneck[9]/ReLU[relu] %1542 : Dynamic = prim::Undefined(), scope: ResNet/Sequential[layer3]/Bottleneck[9]/Conv2d[conv2] %1551 : Float(1, 256, 14, 14) = aten::_convolution[stride=[1, 1], padding=[1, 1], dilation=[1, 1], transposed=0, output_padding=[0, 0], groups=1, benchmark=0, deterministic=0, cudnn_enabled=1](%1541, %319, %1542), scope: ResNet/Sequential[layer3]/Bottleneck[9]/Conv2d[conv2] %1556 : Float(1, 256, 14, 14) = aten::batch_norm[training=0, momentum=0, eps=1e-05, cudnn_enabled=1](%1551, %320, %321, %322, %323), scope: ResNet/Sequential[layer3]/Bottleneck[9]/BatchNorm2d[bn2] %1558 : Float(1, 256, 14, 14) = aten::thresholdthreshold={0}, value={0}, scope: ResNet/Sequential[layer3]/Bottleneck[9]/ReLU[relu] %1559 : Dynamic = prim::Undefined(), scope: ResNet/Sequential[layer3]/Bottleneck[9]/Conv2d[conv3] %1568 : Float(1, 1024, 14, 14) = aten::_convolution[stride=[1, 1], padding=[0, 0], dilation=[1, 1], transposed=0, output_padding=[0, 0], groups=1, benchmark=0, deterministic=0, cudnn_enabled=1](%1558, %325, %1559), scope: ResNet/Sequential[layer3]/Bottleneck[9]/Conv2d[conv3] %1573 : Float(1, 1024, 14, 14) = aten::batch_norm[training=0, momentum=0, eps=1e-05, cudnn_enabled=1](%1568, %326, %327, %328, %329), scope: ResNet/Sequential[layer3]/Bottleneck[9]/BatchNorm2d[bn3] %1574 : Float(1, 1024, 14, 14) = aten::add[alpha={1}](%1573, %1524), scope: ResNet/Sequential[layer3]/Bottleneck[9] %1576 : Float(1, 1024, 14, 14) = aten::thresholdthreshold={0}, value={0}, scope: ResNet/Sequential[layer3]/Bottleneck[9]/ReLU[relu] %1577 : Dynamic = prim::Undefined(), scope: ResNet/Sequential[layer3]/Bottleneck[10]/Conv2d[conv1] %1586 : Float(1, 256, 14, 14) = aten::_convolution[stride=[1, 1], padding=[0, 0], dilation=[1, 1], transposed=0, output_padding=[0, 0], groups=1, benchmark=0, deterministic=0, cudnn_enabled=1](%1576, %331, %1577), scope: ResNet/Sequential[layer3]/Bottleneck[10]/Conv2d[conv1] %1591 : Float(1, 256, 14, 14) = aten::batch_norm[training=0, momentum=0, eps=1e-05, cudnn_enabled=1](%1586, %332, %333, %334, %335), scope: ResNet/Sequential[layer3]/Bottleneck[10]/BatchNorm2d[bn1] %1593 : Float(1, 256, 14, 14) = aten::thresholdthreshold={0}, value={0}, scope: ResNet/Sequential[layer3]/Bottleneck[10]/ReLU[relu] %1594 : Dynamic = prim::Undefined(), scope: ResNet/Sequential[layer3]/Bottleneck[10]/Conv2d[conv2] %1603 : Float(1, 256, 14, 14) = aten::_convolution[stride=[1, 1], padding=[1, 1], dilation=[1, 1], transposed=0, output_padding=[0, 0], groups=1, benchmark=0, deterministic=0, cudnn_enabled=1](%1593, %337, %1594), scope: ResNet/Sequential[layer3]/Bottleneck[10]/Conv2d[conv2] %1608 : Float(1, 256, 14, 14) = aten::batch_norm[training=0, momentum=0, eps=1e-05, cudnn_enabled=1](%1603, %338, %339, %340, %341), scope: ResNet/Sequential[layer3]/Bottleneck[10]/BatchNorm2d[bn2] %1610 : Float(1, 256, 14, 14) = aten::thresholdthreshold={0}, value={0}, scope: ResNet/Sequential[layer3]/Bottleneck[10]/ReLU[relu] %1611 : Dynamic = prim::Undefined(), scope: ResNet/Sequential[layer3]/Bottleneck[10]/Conv2d[conv3] %1620 : Float(1, 1024, 14, 14) = aten::_convolution[stride=[1, 1], padding=[0, 0], dilation=[1, 1], transposed=0, output_padding=[0, 0], groups=1, benchmark=0, deterministic=0, cudnn_enabled=1](%1610, %343, %1611), scope: ResNet/Sequential[layer3]/Bottleneck[10]/Conv2d[conv3] %1625 : Float(1, 1024, 14, 14) = aten::batch_norm[training=0, momentum=0, eps=1e-05, cudnn_enabled=1](%1620, %344, %345, %346, %347), scope: ResNet/Sequential[layer3]/Bottleneck[10]/BatchNorm2d[bn3] %1626 : Float(1, 1024, 14, 14) = aten::add[alpha={1}](%1625, %1576), scope: ResNet/Sequential[layer3]/Bottleneck[10] %1628 : Float(1, 1024, 14, 14) = aten::thresholdthreshold={0}, value={0}, scope: ResNet/Sequential[layer3]/Bottleneck[10]/ReLU[relu] %1629 : Dynamic = prim::Undefined(), scope: ResNet/Sequential[layer3]/Bottleneck[11]/Conv2d[conv1] %1638 : Float(1, 256, 14, 14) = aten::_convolution[stride=[1, 1], padding=[0, 0], dilation=[1, 1], transposed=0, output_padding=[0, 0], groups=1, benchmark=0, deterministic=0, cudnn_enabled=1](%1628, %349, %1629), scope: ResNet/Sequential[layer3]/Bottleneck[11]/Conv2d[conv1] %1643 : Float(1, 256, 14, 14) = aten::batch_norm[training=0, momentum=0, eps=1e-05, cudnn_enabled=1](%1638, %350, %351, %352, %353), scope: ResNet/Sequential[layer3]/Bottleneck[11]/BatchNorm2d[bn1] %1645 : Float(1, 256, 14, 14) = aten::thresholdthreshold={0}, value={0}, scope: ResNet/Sequential[layer3]/Bottleneck[11]/ReLU[relu] %1646 : Dynamic = prim::Undefined(), scope: ResNet/Sequential[layer3]/Bottleneck[11]/Conv2d[conv2] %1655 : Float(1, 256, 14, 14) = aten::_convolution[stride=[1, 1], padding=[1, 1], dilation=[1, 1], transposed=0, output_padding=[0, 0], groups=1, benchmark=0, deterministic=0, cudnn_enabled=1](%1645, %355, %1646), scope: ResNet/Sequential[layer3]/Bottleneck[11]/Conv2d[conv2] %1660 : Float(1, 256, 14, 14) = aten::batch_norm[training=0, momentum=0, eps=1e-05, cudnn_enabled=1](%1655, %356, %357, %358, %359), scope: ResNet/Sequential[layer3]/Bottleneck[11]/BatchNorm2d[bn2] %1662 : Float(1, 256, 14, 14) = aten::thresholdthreshold={0}, value={0}, scope: ResNet/Sequential[layer3]/Bottleneck[11]/ReLU[relu] %1663 : Dynamic = prim::Undefined(), scope: ResNet/Sequential[layer3]/Bottleneck[11]/Conv2d[conv3] %1672 : Float(1, 1024, 14, 14) = aten::_convolution[stride=[1, 1], padding=[0, 0], dilation=[1, 1], transposed=0, output_padding=[0, 0], groups=1, benchmark=0, deterministic=0, cudnn_enabled=1](%1662, %361, %1663), scope: ResNet/Sequential[layer3]/Bottleneck[11]/Conv2d[conv3] %1677 : Float(1, 1024, 14, 14) = aten::batch_norm[training=0, momentum=0, eps=1e-05, cudnn_enabled=1](%1672, %362, %363, %364, %365), scope: ResNet/Sequential[layer3]/Bottleneck[11]/BatchNorm2d[bn3] %1678 : Float(1, 1024, 14, 14) = aten::add[alpha={1}](%1677, %1628), scope: ResNet/Sequential[layer3]/Bottleneck[11] %1680 : Float(1, 1024, 14, 14) = aten::thresholdthreshold={0}, value={0}, scope: ResNet/Sequential[layer3]/Bottleneck[11]/ReLU[relu] %1681 : Dynamic = prim::Undefined(), scope: ResNet/Sequential[layer3]/Bottleneck[12]/Conv2d[conv1] %1690 : Float(1, 256, 14, 14) = aten::_convolution[stride=[1, 1], padding=[0, 0], dilation=[1, 1], transposed=0, output_padding=[0, 0], groups=1, benchmark=0, deterministic=0, cudnn_enabled=1](%1680, %367, %1681), scope: ResNet/Sequential[layer3]/Bottleneck[12]/Conv2d[conv1] %1695 : Float(1, 256, 14, 14) = aten::batch_norm[training=0, momentum=0, eps=1e-05, cudnn_enabled=1](%1690, %368, %369, %370, %371), scope: ResNet/Sequential[layer3]/Bottleneck[12]/BatchNorm2d[bn1] %1697 : Float(1, 256, 14, 14) = aten::thresholdthreshold={0}, value={0}, scope: ResNet/Sequential[layer3]/Bottleneck[12]/ReLU[relu] %1698 : Dynamic = prim::Undefined(), scope: ResNet/Sequential[layer3]/Bottleneck[12]/Conv2d[conv2] %1707 : Float(1, 256, 14, 14) = aten::_convolution[stride=[1, 1], padding=[1, 1], dilation=[1, 1], transposed=0, output_padding=[0, 0], groups=1, benchmark=0, deterministic=0, cudnn_enabled=1](%1697, %373, %1698), scope: ResNet/Sequential[layer3]/Bottleneck[12]/Conv2d[conv2] %1712 : Float(1, 256, 14, 14) = aten::batch_norm[training=0, momentum=0, eps=1e-05, cudnn_enabled=1](%1707, %374, %375, %376, %377), scope: ResNet/Sequential[layer3]/Bottleneck[12]/BatchNorm2d[bn2] %1714 : Float(1, 256, 14, 14) = aten::thresholdthreshold={0}, value={0}, scope: ResNet/Sequential[layer3]/Bottleneck[12]/ReLU[relu] %1715 : Dynamic = prim::Undefined(), scope: ResNet/Sequential[layer3]/Bottleneck[12]/Conv2d[conv3] %1724 : Float(1, 1024, 14, 14) = aten::_convolution[stride=[1, 1], padding=[0, 0], dilation=[1, 1], transposed=0, output_padding=[0, 0], groups=1, benchmark=0, deterministic=0, cudnn_enabled=1](%1714, %379, %1715), scope: ResNet/Sequential[layer3]/Bottleneck[12]/Conv2d[conv3] %1729 : Float(1, 1024, 14, 14) = aten::batch_norm[training=0, momentum=0, eps=1e-05, cudnn_enabled=1](%1724, %380, %381, %382, %383), scope: ResNet/Sequential[layer3]/Bottleneck[12]/BatchNorm2d[bn3] %1730 : Float(1, 1024, 14, 14) = aten::add[alpha={1}](%1729, %1680), scope: ResNet/Sequential[layer3]/Bottleneck[12] %1732 : Float(1, 1024, 14, 14) = aten::thresholdthreshold={0}, value={0}, scope: ResNet/Sequential[layer3]/Bottleneck[12]/ReLU[relu] %1733 : Dynamic = prim::Undefined(), scope: ResNet/Sequential[layer3]/Bottleneck[13]/Conv2d[conv1] %1742 : Float(1, 256, 14, 14) = aten::_convolution[stride=[1, 1], padding=[0, 0], dilation=[1, 1], transposed=0, output_padding=[0, 0], groups=1, benchmark=0, deterministic=0, cudnn_enabled=1](%1732, %385, %1733), scope: ResNet/Sequential[layer3]/Bottleneck[13]/Conv2d[conv1] %1747 : Float(1, 256, 14, 14) = aten::batch_norm[training=0, momentum=0, eps=1e-05, cudnn_enabled=1](%1742, %386, %387, %388, %389), scope: ResNet/Sequential[layer3]/Bottleneck[13]/BatchNorm2d[bn1] %1749 : Float(1, 256, 14, 14) = aten::thresholdthreshold={0}, value={0}, scope: ResNet/Sequential[layer3]/Bottleneck[13]/ReLU[relu] %1750 : Dynamic = prim::Undefined(), scope: ResNet/Sequential[layer3]/Bottleneck[13]/Conv2d[conv2] %1759 : Float(1, 256, 14, 14) = aten::_convolution[stride=[1, 1], padding=[1, 1], dilation=[1, 1], transposed=0, output_padding=[0, 0], groups=1, benchmark=0, deterministic=0, cudnn_enabled=1](%1749, %391, %1750), scope: ResNet/Sequential[layer3]/Bottleneck[13]/Conv2d[conv2] %1764 : Float(1, 256, 14, 14) = aten::batch_norm[training=0, momentum=0, eps=1e-05, cudnn_enabled=1](%1759, %392, %393, %394, %395), scope: ResNet/Sequential[layer3]/Bottleneck[13]/BatchNorm2d[bn2] %1766 : Float(1, 256, 14, 14) = aten::thresholdthreshold={0}, value={0}, scope: ResNet/Sequential[layer3]/Bottleneck[13]/ReLU[relu] %1767 : Dynamic = prim::Undefined(), scope: ResNet/Sequential[layer3]/Bottleneck[13]/Conv2d[conv3] %1776 : Float(1, 1024, 14, 14) = aten::_convolution[stride=[1, 1], padding=[0, 0], dilation=[1, 1], transposed=0, output_padding=[0, 0], groups=1, benchmark=0, deterministic=0, cudnn_enabled=1](%1766, %397, %1767), scope: ResNet/Sequential[layer3]/Bottleneck[13]/Conv2d[conv3] %1781 : Float(1, 1024, 14, 14) = aten::batch_norm[training=0, momentum=0, eps=1e-05, cudnn_enabled=1](%1776, %398, %399, %400, %401), scope: ResNet/Sequential[layer3]/Bottleneck[13]/BatchNorm2d[bn3] %1782 : Float(1, 1024, 14, 14) = aten::add[alpha={1}](%1781, %1732), scope: ResNet/Sequential[layer3]/Bottleneck[13] %1784 : Float(1, 1024, 14, 14) = aten::thresholdthreshold={0}, value={0}, scope: ResNet/Sequential[layer3]/Bottleneck[13]/ReLU[relu] %1785 : Dynamic = prim::Undefined(), scope: ResNet/Sequential[layer3]/Bottleneck[14]/Conv2d[conv1] %1794 : Float(1, 256, 14, 14) = aten::_convolution[stride=[1, 1], padding=[0, 0], dilation=[1, 1], transposed=0, output_padding=[0, 0], groups=1, benchmark=0, deterministic=0, cudnn_enabled=1](%1784, %403, %1785), scope: ResNet/Sequential[layer3]/Bottleneck[14]/Conv2d[conv1] %1799 : Float(1, 256, 14, 14) = aten::batch_norm[training=0, momentum=0, eps=1e-05, cudnn_enabled=1](%1794, %404, %405, %406, %407), scope: ResNet/Sequential[layer3]/Bottleneck[14]/BatchNorm2d[bn1] %1801 : Float(1, 256, 14, 14) = aten::thresholdthreshold={0}, value={0}, scope: ResNet/Sequential[layer3]/Bottleneck[14]/ReLU[relu] %1802 : Dynamic = prim::Undefined(), scope: ResNet/Sequential[layer3]/Bottleneck[14]/Conv2d[conv2] %1811 : Float(1, 256, 14, 14) = aten::_convolution[stride=[1, 1], padding=[1, 1], dilation=[1, 1], transposed=0, output_padding=[0, 0], groups=1, benchmark=0, deterministic=0, cudnn_enabled=1](%1801, %409, %1802), scope: ResNet/Sequential[layer3]/Bottleneck[14]/Conv2d[conv2] %1816 : Float(1, 256, 14, 14) = aten::batch_norm[training=0, momentum=0, eps=1e-05, cudnn_enabled=1](%1811, %410, %411, %412, %413), scope: ResNet/Sequential[layer3]/Bottleneck[14]/BatchNorm2d[bn2] %1818 : Float(1, 256, 14, 14) = aten::thresholdthreshold={0}, value={0}, scope: ResNet/Sequential[layer3]/Bottleneck[14]/ReLU[relu] %1819 : Dynamic = prim::Undefined(), scope: ResNet/Sequential[layer3]/Bottleneck[14]/Conv2d[conv3] %1828 : Float(1, 1024, 14, 14) = aten::_convolution[stride=[1, 1], padding=[0, 0], dilation=[1, 1], transposed=0, output_padding=[0, 0], groups=1, benchmark=0, deterministic=0, cudnn_enabled=1](%1818, %415, %1819), scope: ResNet/Sequential[layer3]/Bottleneck[14]/Conv2d[conv3] %1833 : Float(1, 1024, 14, 14) = aten::batch_norm[training=0, momentum=0, eps=1e-05, cudnn_enabled=1](%1828, %416, %417, %418, %419), scope: ResNet/Sequential[layer3]/Bottleneck[14]/BatchNorm2d[bn3] %1834 : Float(1, 1024, 14, 14) = aten::add[alpha={1}](%1833, %1784), scope: ResNet/Sequential[layer3]/Bottleneck[14] %1836 : Float(1, 1024, 14, 14) = aten::thresholdthreshold={0}, value={0}, scope: ResNet/Sequential[layer3]/Bottleneck[14]/ReLU[relu] %1837 : Dynamic = prim::Undefined(), scope: ResNet/Sequential[layer3]/Bottleneck[15]/Conv2d[conv1] %1846 : Float(1, 256, 14, 14) = aten::_convolution[stride=[1, 1], padding=[0, 0], dilation=[1, 1], transposed=0, output_padding=[0, 0], groups=1, benchmark=0, deterministic=0, cudnn_enabled=1](%1836, %421, %1837), scope: ResNet/Sequential[layer3]/Bottleneck[15]/Conv2d[conv1] %1851 : Float(1, 256, 14, 14) = aten::batch_norm[training=0, momentum=0, eps=1e-05, cudnn_enabled=1](%1846, %422, %423, %424, %425), scope: ResNet/Sequential[layer3]/Bottleneck[15]/BatchNorm2d[bn1] %1853 : Float(1, 256, 14, 14) = aten::thresholdthreshold={0}, value={0}, scope: ResNet/Sequential[layer3]/Bottleneck[15]/ReLU[relu] %1854 : Dynamic = prim::Undefined(), scope: ResNet/Sequential[layer3]/Bottleneck[15]/Conv2d[conv2] %1863 : Float(1, 256, 14, 14) = aten::_convolution[stride=[1, 1], padding=[1, 1], dilation=[1, 1], transposed=0, output_padding=[0, 0], groups=1, benchmark=0, deterministic=0, cudnn_enabled=1](%1853, %427, %1854), scope: ResNet/Sequential[layer3]/Bottleneck[15]/Conv2d[conv2] %1868 : Float(1, 256, 14, 14) = aten::batch_norm[training=0, momentum=0, eps=1e-05, cudnn_enabled=1](%1863, %428, %429, %430, %431), scope: ResNet/Sequential[layer3]/Bottleneck[15]/BatchNorm2d[bn2] %1870 : Float(1, 256, 14, 14) = aten::thresholdthreshold={0}, value={0}, scope: ResNet/Sequential[layer3]/Bottleneck[15]/ReLU[relu] %1871 : Dynamic = prim::Undefined(), scope: ResNet/Sequential[layer3]/Bottleneck[15]/Conv2d[conv3] %1880 : Float(1, 1024, 14, 14) = aten::_convolution[stride=[1, 1], padding=[0, 0], dilation=[1, 1], transposed=0, output_padding=[0, 0], groups=1, benchmark=0, deterministic=0, cudnn_enabled=1](%1870, %433, %1871), scope: ResNet/Sequential[layer3]/Bottleneck[15]/Conv2d[conv3] %1885 : Float(1, 1024, 14, 14) = aten::batch_norm[training=0, momentum=0, eps=1e-05, cudnn_enabled=1](%1880, %434, %435, %436, %437), scope: ResNet/Sequential[layer3]/Bottleneck[15]/BatchNorm2d[bn3] %1886 : Float(1, 1024, 14, 14) = aten::add[alpha={1}](%1885, %1836), scope: ResNet/Sequential[layer3]/Bottleneck[15] %1888 : Float(1, 1024, 14, 14) = aten::thresholdthreshold={0}, value={0}, scope: ResNet/Sequential[layer3]/Bottleneck[15]/ReLU[relu] %1889 : Dynamic = prim::Undefined(), scope: ResNet/Sequential[layer3]/Bottleneck[16]/Conv2d[conv1] %1898 : Float(1, 256, 14, 14) = aten::_convolution[stride=[1, 1], padding=[0, 0], dilation=[1, 1], transposed=0, output_padding=[0, 0], groups=1, benchmark=0, deterministic=0, cudnn_enabled=1](%1888, %439, %1889), scope: ResNet/Sequential[layer3]/Bottleneck[16]/Conv2d[conv1] %1903 : Float(1, 256, 14, 14) = aten::batch_norm[training=0, momentum=0, eps=1e-05, cudnn_enabled=1](%1898, %440, %441, %442, %443), scope: ResNet/Sequential[layer3]/Bottleneck[16]/BatchNorm2d[bn1] %1905 : Float(1, 256, 14, 14) = aten::thresholdthreshold={0}, value={0}, scope: ResNet/Sequential[layer3]/Bottleneck[16]/ReLU[relu] %1906 : Dynamic = prim::Undefined(), scope: ResNet/Sequential[layer3]/Bottleneck[16]/Conv2d[conv2] %1915 : Float(1, 256, 14, 14) = aten::_convolution[stride=[1, 1], padding=[1, 1], dilation=[1, 1], transposed=0, output_padding=[0, 0], groups=1, benchmark=0, deterministic=0, cudnn_enabled=1](%1905, %445, %1906), scope: ResNet/Sequential[layer3]/Bottleneck[16]/Conv2d[conv2] %1920 : Float(1, 256, 14, 14) = aten::batch_norm[training=0, momentum=0, eps=1e-05, cudnn_enabled=1](%1915, %446, %447, %448, %449), scope: ResNet/Sequential[layer3]/Bottleneck[16]/BatchNorm2d[bn2] %1922 : Float(1, 256, 14, 14) = aten::thresholdthreshold={0}, value={0}, scope: ResNet/Sequential[layer3]/Bottleneck[16]/ReLU[relu] %1923 : Dynamic = prim::Undefined(), scope: ResNet/Sequential[layer3]/Bottleneck[16]/Conv2d[conv3] %1932 : Float(1, 1024, 14, 14) = aten::_convolution[stride=[1, 1], padding=[0, 0], dilation=[1, 1], transposed=0, output_padding=[0, 0], groups=1, benchmark=0, deterministic=0, cudnn_enabled=1](%1922, %451, %1923), scope: ResNet/Sequential[layer3]/Bottleneck[16]/Conv2d[conv3] %1937 : Float(1, 1024, 14, 14) = aten::batch_norm[training=0, momentum=0, eps=1e-05, cudnn_enabled=1](%1932, %452, %453, %454, %455), scope: ResNet/Sequential[layer3]/Bottleneck[16]/BatchNorm2d[bn3] %1938 : Float(1, 1024, 14, 14) = aten::add[alpha={1}](%1937, %1888), scope: ResNet/Sequential[layer3]/Bottleneck[16] %1940 : Float(1, 1024, 14, 14) = aten::thresholdthreshold={0}, value={0}, scope: ResNet/Sequential[layer3]/Bottleneck[16]/ReLU[relu] %1941 : Dynamic = prim::Undefined(), scope: ResNet/Sequential[layer3]/Bottleneck[17]/Conv2d[conv1] %1950 : Float(1, 256, 14, 14) = aten::_convolution[stride=[1, 1], padding=[0, 0], dilation=[1, 1], transposed=0, output_padding=[0, 0], groups=1, benchmark=0, deterministic=0, cudnn_enabled=1](%1940, %457, %1941), scope: ResNet/Sequential[layer3]/Bottleneck[17]/Conv2d[conv1] %1955 : Float(1, 256, 14, 14) = aten::batch_norm[training=0, momentum=0, eps=1e-05, cudnn_enabled=1](%1950, %458, %459, %460, %461), scope: ResNet/Sequential[layer3]/Bottleneck[17]/BatchNorm2d[bn1] %1957 : Float(1, 256, 14, 14) = aten::thresholdthreshold={0}, value={0}, scope: ResNet/Sequential[layer3]/Bottleneck[17]/ReLU[relu] %1958 : Dynamic = prim::Undefined(), scope: ResNet/Sequential[layer3]/Bottleneck[17]/Conv2d[conv2] %1967 : Float(1, 256, 14, 14) = aten::_convolution[stride=[1, 1], padding=[1, 1], dilation=[1, 1], transposed=0, output_padding=[0, 0], groups=1, benchmark=0, deterministic=0, cudnn_enabled=1](%1957, %463, %1958), scope: ResNet/Sequential[layer3]/Bottleneck[17]/Conv2d[conv2] %1972 : Float(1, 256, 14, 14) = aten::batch_norm[training=0, momentum=0, eps=1e-05, cudnn_enabled=1](%1967, %464, %465, %466, %467), scope: ResNet/Sequential[layer3]/Bottleneck[17]/BatchNorm2d[bn2] %1974 : Float(1, 256, 14, 14) = aten::thresholdthreshold={0}, value={0}, scope: ResNet/Sequential[layer3]/Bottleneck[17]/ReLU[relu] %1975 : Dynamic = prim::Undefined(), scope: ResNet/Sequential[layer3]/Bottleneck[17]/Conv2d[conv3] %1984 : Float(1, 1024, 14, 14) = aten::_convolution[stride=[1, 1], padding=[0, 0], dilation=[1, 1], transposed=0, output_padding=[0, 0], groups=1, benchmark=0, deterministic=0, cudnn_enabled=1](%1974, %469, %1975), scope: ResNet/Sequential[layer3]/Bottleneck[17]/Conv2d[conv3] %1989 : Float(1, 1024, 14, 14) = aten::batch_norm[training=0, momentum=0, eps=1e-05, cudnn_enabled=1](%1984, %470, %471, %472, %473), scope: ResNet/Sequential[layer3]/Bottleneck[17]/BatchNorm2d[bn3] %1990 : Float(1, 1024, 14, 14) = aten::add[alpha={1}](%1989, %1940), scope: ResNet/Sequential[layer3]/Bottleneck[17] %1992 : Float(1, 1024, 14, 14) = aten::thresholdthreshold={0}, value={0}, scope: ResNet/Sequential[layer3]/Bottleneck[17]/ReLU[relu] %1993 : Dynamic = prim::Undefined(), scope: ResNet/Sequential[layer3]/Bottleneck[18]/Conv2d[conv1] %2002 : Float(1, 256, 14, 14) = aten::_convolution[stride=[1, 1], padding=[0, 0], dilation=[1, 1], transposed=0, output_padding=[0, 0], groups=1, benchmark=0, deterministic=0, cudnn_enabled=1](%1992, %475, %1993), scope: ResNet/Sequential[layer3]/Bottleneck[18]/Conv2d[conv1] %2007 : Float(1, 256, 14, 14) = aten::batch_norm[training=0, momentum=0, eps=1e-05, cudnn_enabled=1](%2002, %476, %477, %478, %479), scope: ResNet/Sequential[layer3]/Bottleneck[18]/BatchNorm2d[bn1] %2009 : Float(1, 256, 14, 14) = aten::thresholdthreshold={0}, value={0}, scope: ResNet/Sequential[layer3]/Bottleneck[18]/ReLU[relu] %2010 : Dynamic = prim::Undefined(), scope: ResNet/Sequential[layer3]/Bottleneck[18]/Conv2d[conv2] %2019 : Float(1, 256, 14, 14) = aten::_convolution[stride=[1, 1], padding=[1, 1], dilation=[1, 1], transposed=0, output_padding=[0, 0], groups=1, benchmark=0, deterministic=0, cudnn_enabled=1](%2009, %481, %2010), scope: ResNet/Sequential[layer3]/Bottleneck[18]/Conv2d[conv2] %2024 : Float(1, 256, 14, 14) = aten::batch_norm[training=0, momentum=0, eps=1e-05, cudnn_enabled=1](%2019, %482, %483, %484, %485), scope: ResNet/Sequential[layer3]/Bottleneck[18]/BatchNorm2d[bn2] %2026 : Float(1, 256, 14, 14) = aten::thresholdthreshold={0}, value={0}, scope: ResNet/Sequential[layer3]/Bottleneck[18]/ReLU[relu] %2027 : Dynamic = prim::Undefined(), scope: ResNet/Sequential[layer3]/Bottleneck[18]/Conv2d[conv3] %2036 : Float(1, 1024, 14, 14) = aten::_convolution[stride=[1, 1], padding=[0, 0], dilation=[1, 1], transposed=0, output_padding=[0, 0], groups=1, benchmark=0, deterministic=0, cudnn_enabled=1](%2026, %487, %2027), scope: ResNet/Sequential[layer3]/Bottleneck[18]/Conv2d[conv3] %2041 : Float(1, 1024, 14, 14) = aten::batch_norm[training=0, momentum=0, eps=1e-05, cudnn_enabled=1](%2036, %488, %489, %490, %491), scope: ResNet/Sequential[layer3]/Bottleneck[18]/BatchNorm2d[bn3] %2042 : Float(1, 1024, 14, 14) = aten::add[alpha={1}](%2041, %1992), scope: ResNet/Sequential[layer3]/Bottleneck[18] %2044 : Float(1, 1024, 14, 14) = aten::thresholdthreshold={0}, value={0}, scope: ResNet/Sequential[layer3]/Bottleneck[18]/ReLU[relu] %2045 : Dynamic = prim::Undefined(), scope: ResNet/Sequential[layer3]/Bottleneck[19]/Conv2d[conv1] %2054 : Float(1, 256, 14, 14) = aten::_convolution[stride=[1, 1], padding=[0, 0], dilation=[1, 1], transposed=0, output_padding=[0, 0], groups=1, benchmark=0, deterministic=0, cudnn_enabled=1](%2044, %493, %2045), scope: ResNet/Sequential[layer3]/Bottleneck[19]/Conv2d[conv1] %2059 : Float(1, 256, 14, 14) = aten::batch_norm[training=0, momentum=0, eps=1e-05, cudnn_enabled=1](%2054, %494, %495, %496, %497), scope: ResNet/Sequential[layer3]/Bottleneck[19]/BatchNorm2d[bn1] %2061 : Float(1, 256, 14, 14) = aten::thresholdthreshold={0}, value={0}, scope: ResNet/Sequential[layer3]/Bottleneck[19]/ReLU[relu] %2062 : Dynamic = prim::Undefined(), scope: ResNet/Sequential[layer3]/Bottleneck[19]/Conv2d[conv2] %2071 : Float(1, 256, 14, 14) = aten::_convolution[stride=[1, 1], padding=[1, 1], dilation=[1, 1], transposed=0, output_padding=[0, 0], groups=1, benchmark=0, deterministic=0, cudnn_enabled=1](%2061, %499, %2062), scope: ResNet/Sequential[layer3]/Bottleneck[19]/Conv2d[conv2] %2076 : Float(1, 256, 14, 14) = aten::batch_norm[training=0, momentum=0, eps=1e-05, cudnn_enabled=1](%2071, %500, %501, %502, %503), scope: ResNet/Sequential[layer3]/Bottleneck[19]/BatchNorm2d[bn2] %2078 : Float(1, 256, 14, 14) = aten::thresholdthreshold={0}, value={0}, scope: ResNet/Sequential[layer3]/Bottleneck[19]/ReLU[relu] %2079 : Dynamic = prim::Undefined(), scope: ResNet/Sequential[layer3]/Bottleneck[19]/Conv2d[conv3] %2088 : Float(1, 1024, 14, 14) = aten::_convolution[stride=[1, 1], padding=[0, 0], dilation=[1, 1], transposed=0, output_padding=[0, 0], groups=1, benchmark=0, deterministic=0, cudnn_enabled=1](%2078, %505, %2079), scope: ResNet/Sequential[layer3]/Bottleneck[19]/Conv2d[conv3] %2093 : Float(1, 1024, 14, 14) = aten::batch_norm[training=0, momentum=0, eps=1e-05, cudnn_enabled=1](%2088, %506, %507, %508, %509), scope: ResNet/Sequential[layer3]/Bottleneck[19]/BatchNorm2d[bn3] %2094 : Float(1, 1024, 14, 14) = aten::add[alpha={1}](%2093, %2044), scope: ResNet/Sequential[layer3]/Bottleneck[19] %2096 : Float(1, 1024, 14, 14) = aten::thresholdthreshold={0}, value={0}, scope: ResNet/Sequential[layer3]/Bottleneck[19]/ReLU[relu] %2097 : Dynamic = prim::Undefined(), scope: ResNet/Sequential[layer3]/Bottleneck[20]/Conv2d[conv1] %2106 : Float(1, 256, 14, 14) = aten::_convolution[stride=[1, 1], padding=[0, 0], dilation=[1, 1], transposed=0, output_padding=[0, 0], groups=1, benchmark=0, deterministic=0, cudnn_enabled=1](%2096, %511, %2097), scope: ResNet/Sequential[layer3]/Bottleneck[20]/Conv2d[conv1] %2111 : Float(1, 256, 14, 14) = aten::batch_norm[training=0, momentum=0, eps=1e-05, cudnn_enabled=1](%2106, %512, %513, %514, %515), scope: ResNet/Sequential[layer3]/Bottleneck[20]/BatchNorm2d[bn1] %2113 : Float(1, 256, 14, 14) = aten::thresholdthreshold={0}, value={0}, scope: ResNet/Sequential[layer3]/Bottleneck[20]/ReLU[relu] %2114 : Dynamic = prim::Undefined(), scope: ResNet/Sequential[layer3]/Bottleneck[20]/Conv2d[conv2] %2123 : Float(1, 256, 14, 14) = aten::_convolution[stride=[1, 1], padding=[1, 1], dilation=[1, 1], transposed=0, output_padding=[0, 0], groups=1, benchmark=0, deterministic=0, cudnn_enabled=1](%2113, %517, %2114), scope: ResNet/Sequential[layer3]/Bottleneck[20]/Conv2d[conv2] %2128 : Float(1, 256, 14, 14) = aten::batch_norm[training=0, momentum=0, eps=1e-05, cudnn_enabled=1](%2123, %518, %519, %520, %521), scope: ResNet/Sequential[layer3]/Bottleneck[20]/BatchNorm2d[bn2] %2130 : Float(1, 256, 14, 14) = aten::thresholdthreshold={0}, value={0}, scope: ResNet/Sequential[layer3]/Bottleneck[20]/ReLU[relu] %2131 : Dynamic = prim::Undefined(), scope: ResNet/Sequential[layer3]/Bottleneck[20]/Conv2d[conv3] %2140 : Float(1, 1024, 14, 14) = aten::_convolution[stride=[1, 1], padding=[0, 0], dilation=[1, 1], transposed=0, output_padding=[0, 0], groups=1, benchmark=0, deterministic=0, cudnn_enabled=1](%2130, %523, %2131), scope: ResNet/Sequential[layer3]/Bottleneck[20]/Conv2d[conv3] %2145 : Float(1, 1024, 14, 14) = aten::batch_norm[training=0, momentum=0, eps=1e-05, cudnn_enabled=1](%2140, %524, %525, %526, %527), scope: ResNet/Sequential[layer3]/Bottleneck[20]/BatchNorm2d[bn3] %2146 : Float(1, 1024, 14, 14) = aten::add[alpha={1}](%2145, %2096), scope: ResNet/Sequential[layer3]/Bottleneck[20] %2148 : Float(1, 1024, 14, 14) = aten::thresholdthreshold={0}, value={0}, scope: ResNet/Sequential[layer3]/Bottleneck[20]/ReLU[relu] %2149 : Dynamic = prim::Undefined(), scope: ResNet/Sequential[layer3]/Bottleneck[21]/Conv2d[conv1] %2158 : Float(1, 256, 14, 14) = aten::_convolution[stride=[1, 1], padding=[0, 0], dilation=[1, 1], transposed=0, output_padding=[0, 0], groups=1, benchmark=0, deterministic=0, cudnn_enabled=1](%2148, %529, %2149), scope: ResNet/Sequential[layer3]/Bottleneck[21]/Conv2d[conv1] %2163 : Float(1, 256, 14, 14) = aten::batch_norm[training=0, momentum=0, eps=1e-05, cudnn_enabled=1](%2158, %530, %531, %532, %533), scope: ResNet/Sequential[layer3]/Bottleneck[21]/BatchNorm2d[bn1] %2165 : Float(1, 256, 14, 14) = aten::thresholdthreshold={0}, value={0}, scope: ResNet/Sequential[layer3]/Bottleneck[21]/ReLU[relu] %2166 : Dynamic = prim::Undefined(), scope: ResNet/Sequential[layer3]/Bottleneck[21]/Conv2d[conv2] %2175 : Float(1, 256, 14, 14) = aten::_convolution[stride=[1, 1], padding=[1, 1], dilation=[1, 1], transposed=0, output_padding=[0, 0], groups=1, benchmark=0, deterministic=0, cudnn_enabled=1](%2165, %535, %2166), scope: ResNet/Sequential[layer3]/Bottleneck[21]/Conv2d[conv2] %2180 : Float(1, 256, 14, 14) = aten::batch_norm[training=0, momentum=0, eps=1e-05, cudnn_enabled=1](%2175, %536, %537, %538, %539), scope: ResNet/Sequential[layer3]/Bottleneck[21]/BatchNorm2d[bn2] %2182 : Float(1, 256, 14, 14) = aten::thresholdthreshold={0}, value={0}, scope: ResNet/Sequential[layer3]/Bottleneck[21]/ReLU[relu] %2183 : Dynamic = prim::Undefined(), scope: ResNet/Sequential[layer3]/Bottleneck[21]/Conv2d[conv3] %2192 : Float(1, 1024, 14, 14) = aten::_convolution[stride=[1, 1], padding=[0, 0], dilation=[1, 1], transposed=0, output_padding=[0, 0], groups=1, benchmark=0, deterministic=0, cudnn_enabled=1](%2182, %541, %2183), scope: ResNet/Sequential[layer3]/Bottleneck[21]/Conv2d[conv3] %2197 : Float(1, 1024, 14, 14) = aten::batch_norm[training=0, momentum=0, eps=1e-05, cudnn_enabled=1](%2192, %542, %543, %544, %545), scope: ResNet/Sequential[layer3]/Bottleneck[21]/BatchNorm2d[bn3] %2198 : Float(1, 1024, 14, 14) = aten::add[alpha={1}](%2197, %2148), scope: ResNet/Sequential[layer3]/Bottleneck[21] %2200 : Float(1, 1024, 14, 14) = aten::thresholdthreshold={0}, value={0}, scope: ResNet/Sequential[layer3]/Bottleneck[21]/ReLU[relu] %2201 : Dynamic = prim::Undefined(), scope: ResNet/Sequential[layer3]/Bottleneck[22]/Conv2d[conv1] %2210 : Float(1, 256, 14, 14) = aten::_convolution[stride=[1, 1], padding=[0, 0], dilation=[1, 1], transposed=0, output_padding=[0, 0], groups=1, benchmark=0, deterministic=0, cudnn_enabled=1](%2200, %547, %2201), scope: ResNet/Sequential[layer3]/Bottleneck[22]/Conv2d[conv1] %2215 : Float(1, 256, 14, 14) = aten::batch_norm[training=0, momentum=0, eps=1e-05, cudnn_enabled=1](%2210, %548, %549, %550, %551), scope: ResNet/Sequential[layer3]/Bottleneck[22]/BatchNorm2d[bn1] %2217 : Float(1, 256, 14, 14) = aten::thresholdthreshold={0}, value={0}, scope: ResNet/Sequential[layer3]/Bottleneck[22]/ReLU[relu] %2218 : Dynamic = prim::Undefined(), scope: ResNet/Sequential[layer3]/Bottleneck[22]/Conv2d[conv2] %2227 : Float(1, 256, 14, 14) = aten::_convolution[stride=[1, 1], padding=[1, 1], dilation=[1, 1], transposed=0, output_padding=[0, 0], groups=1, benchmark=0, deterministic=0, cudnn_enabled=1](%2217, %553, %2218), scope: ResNet/Sequential[layer3]/Bottleneck[22]/Conv2d[conv2] %2232 : Float(1, 256, 14, 14) = aten::batch_norm[training=0, momentum=0, eps=1e-05, cudnn_enabled=1](%2227, %554, %555, %556, %557), scope: ResNet/Sequential[layer3]/Bottleneck[22]/BatchNorm2d[bn2] %2234 : Float(1, 256, 14, 14) = aten::thresholdthreshold={0}, value={0}, scope: ResNet/Sequential[layer3]/Bottleneck[22]/ReLU[relu] %2235 : Dynamic = prim::Undefined(), scope: ResNet/Sequential[layer3]/Bottleneck[22]/Conv2d[conv3] %2244 : Float(1, 1024, 14, 14) = aten::_convolution[stride=[1, 1], padding=[0, 0], dilation=[1, 1], transposed=0, output_padding=[0, 0], groups=1, benchmark=0, deterministic=0, cudnn_enabled=1](%2234, %559, %2235), scope: ResNet/Sequential[layer3]/Bottleneck[22]/Conv2d[conv3] %2249 : Float(1, 1024, 14, 14) = aten::batch_norm[training=0, momentum=0, eps=1e-05, cudnn_enabled=1](%2244, %560, %561, %562, %563), scope: ResNet/Sequential[layer3]/Bottleneck[22]/BatchNorm2d[bn3] %2250 : Float(1, 1024, 14, 14) = aten::add[alpha={1}](%2249, %2200), scope: ResNet/Sequential[layer3]/Bottleneck[22] %2252 : Float(1, 1024, 14, 14) = aten::thresholdthreshold={0}, value={0}, scope: ResNet/Sequential[layer3]/Bottleneck[22]/ReLU[relu] %2253 : Dynamic = prim::Undefined(), scope: ResNet/Sequential[layer4]/Bottleneck[0]/Conv2d[conv1] %2262 : Float(1, 512, 14, 14) = aten::_convolution[stride=[1, 1], padding=[0, 0], dilation=[1, 1], transposed=0, output_padding=[0, 0], groups=1, benchmark=0, deterministic=0, cudnn_enabled=1](%2252, %565, %2253), scope: ResNet/Sequential[layer4]/Bottleneck[0]/Conv2d[conv1] %2267 : Float(1, 512, 14, 14) = aten::batch_norm[training=0, momentum=0, eps=1e-05, cudnn_enabled=1](%2262, %566, %567, %568, %569), scope: ResNet/Sequential[layer4]/Bottleneck[0]/BatchNorm2d[bn1] %2269 : Float(1, 512, 14, 14) = aten::thresholdthreshold={0}, value={0}, scope: ResNet/Sequential[layer4]/Bottleneck[0]/ReLU[relu] %2270 : Dynamic = prim::Undefined(), scope: ResNet/Sequential[layer4]/Bottleneck[0]/Conv2d[conv2] %2279 : Float(1, 512, 7, 7) = aten::_convolution[stride=[2, 2], padding=[1, 1], dilation=[1, 1], transposed=0, output_padding=[0, 0], groups=1, benchmark=0, deterministic=0, cudnn_enabled=1](%2269, %571, %2270), scope: ResNet/Sequential[layer4]/Bottleneck[0]/Conv2d[conv2] %2284 : Float(1, 512, 7, 7) = aten::batch_norm[training=0, momentum=0, eps=1e-05, cudnn_enabled=1](%2279, %572, %573, %574, %575), scope: ResNet/Sequential[layer4]/Bottleneck[0]/BatchNorm2d[bn2] %2286 : Float(1, 512, 7, 7) = aten::thresholdthreshold={0}, value={0}, scope: ResNet/Sequential[layer4]/Bottleneck[0]/ReLU[relu] %2287 : Dynamic = prim::Undefined(), scope: ResNet/Sequential[layer4]/Bottleneck[0]/Conv2d[conv3] %2296 : Float(1, 2048, 7, 7) = aten::_convolution[stride=[1, 1], padding=[0, 0], dilation=[1, 1], transposed=0, output_padding=[0, 0], groups=1, benchmark=0, deterministic=0, cudnn_enabled=1](%2286, %577, %2287), scope: ResNet/Sequential[layer4]/Bottleneck[0]/Conv2d[conv3] %2301 : Float(1, 2048, 7, 7) = aten::batch_norm[training=0, momentum=0, eps=1e-05, cudnn_enabled=1](%2296, %578, %579, %580, %581), scope: ResNet/Sequential[layer4]/Bottleneck[0]/BatchNorm2d[bn3] %2302 : Dynamic = prim::Undefined(), scope: ResNet/Sequential[layer4]/Bottleneck[0]/Sequential[downsample]/Conv2d[0] %2311 : Float(1, 2048, 7, 7) = aten::_convolution[stride=[2, 2], padding=[0, 0], dilation=[1, 1], transposed=0, output_padding=[0, 0], groups=1, benchmark=0, deterministic=0, cudnn_enabled=1](%2252, %583, %2302), scope: ResNet/Sequential[layer4]/Bottleneck[0]/Sequential[downsample]/Conv2d[0] %2316 : Float(1, 2048, 7, 7) = aten::batch_norm[training=0, momentum=0, eps=1e-05, cudnn_enabled=1](%2311, %584, %585, %586, %587), scope: ResNet/Sequential[layer4]/Bottleneck[0]/Sequential[downsample]/BatchNorm2d[1] %2317 : Float(1, 2048, 7, 7) = aten::add[alpha={1}](%2301, %2316), scope: ResNet/Sequential[layer4]/Bottleneck[0] %2319 : Float(1, 2048, 7, 7) = aten::thresholdthreshold={0}, value={0}, scope: ResNet/Sequential[layer4]/Bottleneck[0]/ReLU[relu] %2320 : Dynamic = prim::Undefined(), scope: ResNet/Sequential[layer4]/Bottleneck[1]/Conv2d[conv1] %2329 : Float(1, 512, 7, 7) = aten::_convolution[stride=[1, 1], padding=[0, 0], dilation=[1, 1], transposed=0, output_padding=[0, 0], groups=1, benchmark=0, deterministic=0, cudnn_enabled=1](%2319, %589, %2320), scope: ResNet/Sequential[layer4]/Bottleneck[1]/Conv2d[conv1] %2334 : Float(1, 512, 7, 7) = aten::batch_norm[training=0, momentum=0, eps=1e-05, cudnn_enabled=1](%2329, %590, %591, %592, %593), scope: ResNet/Sequential[layer4]/Bottleneck[1]/BatchNorm2d[bn1] %2336 : Float(1, 512, 7, 7) = aten::thresholdthreshold={0}, value={0}, scope: ResNet/Sequential[layer4]/Bottleneck[1]/ReLU[relu] %2337 : Dynamic = prim::Undefined(), scope: ResNet/Sequential[layer4]/Bottleneck[1]/Conv2d[conv2] %2346 : Float(1, 512, 7, 7) = aten::_convolution[stride=[1, 1], padding=[1, 1], dilation=[1, 1], transposed=0, output_padding=[0, 0], groups=1, benchmark=0, deterministic=0, cudnn_enabled=1](%2336, %595, %2337), scope: ResNet/Sequential[layer4]/Bottleneck[1]/Conv2d[conv2] %2351 : Float(1, 512, 7, 7) = aten::batch_norm[training=0, momentum=0, eps=1e-05, cudnn_enabled=1](%2346, %596, %597, %598, %599), scope: ResNet/Sequential[layer4]/Bottleneck[1]/BatchNorm2d[bn2] %2353 : Float(1, 512, 7, 7) = aten::thresholdthreshold={0}, value={0}, scope: ResNet/Sequential[layer4]/Bottleneck[1]/ReLU[relu] %2354 : Dynamic = prim::Undefined(), scope: ResNet/Sequential[layer4]/Bottleneck[1]/Conv2d[conv3] %2363 : Float(1, 2048, 7, 7) = aten::_convolution[stride=[1, 1], padding=[0, 0], dilation=[1, 1], transposed=0, output_padding=[0, 0], groups=1, benchmark=0, deterministic=0, cudnn_enabled=1](%2353, %601, %2354), scope: ResNet/Sequential[layer4]/Bottleneck[1]/Conv2d[conv3] %2368 : Float(1, 2048, 7, 7) = aten::batch_norm[training=0, momentum=0, eps=1e-05, cudnn_enabled=1](%2363, %602, %603, %604, %605), scope: ResNet/Sequential[layer4]/Bottleneck[1]/BatchNorm2d[bn3] %2369 : Float(1, 2048, 7, 7) = aten::add[alpha={1}](%2368, %2319), scope: ResNet/Sequential[layer4]/Bottleneck[1] %2371 : Float(1, 2048, 7, 7) = aten::thresholdthreshold={0}, value={0}, scope: ResNet/Sequential[layer4]/Bottleneck[1]/ReLU[relu] %2372 : Dynamic = prim::Undefined(), scope: ResNet/Sequential[layer4]/Bottleneck[2]/Conv2d[conv1] %2381 : Float(1, 512, 7, 7) = aten::_convolution[stride=[1, 1], padding=[0, 0], dilation=[1, 1], transposed=0, output_padding=[0, 0], groups=1, benchmark=0, deterministic=0, cudnn_enabled=1](%2371, %607, %2372), scope: ResNet/Sequential[layer4]/Bottleneck[2]/Conv2d[conv1] %2386 : Float(1, 512, 7, 7) = aten::batch_norm[training=0, momentum=0, eps=1e-05, cudnn_enabled=1](%2381, %608, %609, %610, %611), scope: ResNet/Sequential[layer4]/Bottleneck[2]/BatchNorm2d[bn1] %2388 : Float(1, 512, 7, 7) = aten::thresholdthreshold={0}, value={0}, scope: ResNet/Sequential[layer4]/Bottleneck[2]/ReLU[relu] %2389 : Dynamic = prim::Undefined(), scope: ResNet/Sequential[layer4]/Bottleneck[2]/Conv2d[conv2] %2398 : Float(1, 512, 7, 7) = aten::_convolution[stride=[1, 1], padding=[1, 1], dilation=[1, 1], transposed=0, output_padding=[0, 0], groups=1, benchmark=0, deterministic=0, cudnn_enabled=1](%2388, %613, %2389), scope: ResNet/Sequential[layer4]/Bottleneck[2]/Conv2d[conv2] %2403 : Float(1, 512, 7, 7) = aten::batch_norm[training=0, momentum=0, eps=1e-05, cudnn_enabled=1](%2398, %614, %615, %616, %617), scope: ResNet/Sequential[layer4]/Bottleneck[2]/BatchNorm2d[bn2] %2405 : Float(1, 512, 7, 7) = aten::thresholdthreshold={0}, value={0}, scope: ResNet/Sequential[layer4]/Bottleneck[2]/ReLU[relu] %2406 : Dynamic = prim::Undefined(), scope: ResNet/Sequential[layer4]/Bottleneck[2]/Conv2d[conv3] %2415 : Float(1, 2048, 7, 7) = aten::_convolution[stride=[1, 1], padding=[0, 0], dilation=[1, 1], transposed=0, output_padding=[0, 0], groups=1, benchmark=0, deterministic=0, cudnn_enabled=1](%2405, %619, %2406), scope: ResNet/Sequential[layer4]/Bottleneck[2]/Conv2d[conv3] %2420 : Float(1, 2048, 7, 7) = aten::batch_norm[training=0, momentum=0, eps=1e-05, cudnn_enabled=1](%2415, %620, %621, %622, %623), scope: ResNet/Sequential[layer4]/Bottleneck[2]/BatchNorm2d[bn3] %2421 : Float(1, 2048, 7, 7) = aten::add[alpha={1}](%2420, %2371), scope: ResNet/Sequential[layer4]/Bottleneck[2] %2423 : Float(1, 2048, 7, 7) = aten::thresholdthreshold={0}, value={0}, scope: ResNet/Sequential[layer4]/Bottleneck[2]/ReLU[relu] %2425 : Float(1, 2048, 1, 1) = aten::avg_pool2dkernel_size=[7, 7], stride=[1, 1], padding=[0, 0], ceil_mode=0, count_include_pad=1, scope: ResNet/AvgPool2d[avgpool] %2426 : Long() = aten::sizedim=0, scope: ResNet %2427 : Long() = prim::Constant[value={-1}](), scope: ResNet %2428 : Dynamic = aten::stack[dim=0](%2426, %2427), scope: ResNet %2429 : Float(1, 2048) = aten::view(%2425, %2428), scope: ResNet %2430 : Float(2048!, 1000!) = aten::t(%625), scope: ResNet/Linear[fc] %2431 : Float(1, 1000) = aten::expandsize=[1, 1000], implicit=1, scope: ResNet/Linear[fc] %2432 : Float(1, 1000) = aten::addmm[beta={1}, alpha={1}](%2431, %2429, %2430), scope: ResNet/Linear[fc] return (%2432); } , False

kitstar commented 5 years ago

Hi @charlesrwest , the 0.4.0 is only supported pytorch version in current MMdnn. And we will update our tools to support pytorch 1.0 when it is released.

onnx20 commented 5 years ago

Hi @charlesrwest , the 0.4.0 is only supported pytorch version in current MMdnn. And we will update our tools to support pytorch 1.0 when it is released.

Hi,Is the pytorch-1.0.0 supported ? I have the same problem