Tensor RT issue https://github.com/pytorch/TensorRT/issues/1628

gistfile1.txt

orch-TensorRT TorchScript Conversion Context] - CaskConvolution has no valid tactics for this config, skipping
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %input.155 : Tensor = aten::_convolution(%234, %self.kp_detector.fg_encoder.layer2.0.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.0/__module.kp_detector.fg_encoder.layer2.0.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.21 : Tensor = aten::batch_norm(%input.155, %self.kp_detector.fg_encoder.layer2.0.bn2.weight, %self.kp_detector.fg_encoder.layer2.0.bn2.bias, %self.kp_detector.fg_encoder.layer2.0.bn2.running_mean, %self.kp_detector.fg_encoder.layer2.0.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.0/__module.kp_detector.fg_encoder.layer2.0.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 (CaskFlattenConvolution)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - CaskFlattenConvolution has no valid tactics for this config, skipping
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Half(32768,1:4,1024,32) -> Half(32768,1:4,1024,32) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %input.155 : Tensor = aten::_convolution(%234, %self.kp_detector.fg_encoder.layer2.0.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.0/__module.kp_detector.fg_encoder.layer2.0.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.21 : Tensor = aten::batch_norm(%input.155, %self.kp_detector.fg_encoder.layer2.0.bn2.weight, %self.kp_detector.fg_encoder.layer2.0.bn2.bias, %self.kp_detector.fg_encoder.layer2.0.bn2.running_mean, %self.kp_detector.fg_encoder.layer2.0.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.0/__module.kp_detector.fg_encoder.layer2.0.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 (CaskConvolution)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - CaskConvolution has no valid tactics for this config, skipping
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %input.155 : Tensor = aten::_convolution(%234, %self.kp_detector.fg_encoder.layer2.0.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.0/__module.kp_detector.fg_encoder.layer2.0.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.21 : Tensor = aten::batch_norm(%input.155, %self.kp_detector.fg_encoder.layer2.0.bn2.weight, %self.kp_detector.fg_encoder.layer2.0.bn2.bias, %self.kp_detector.fg_encoder.layer2.0.bn2.running_mean, %self.kp_detector.fg_encoder.layer2.0.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.0/__module.kp_detector.fg_encoder.layer2.0.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 (CaskFlattenConvolution)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - CaskFlattenConvolution has no valid tactics for this config, skipping
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Half(16384,1:8,512,16) -> Float(131072,1024,32,1) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %input.155 : Tensor = aten::_convolution(%234, %self.kp_detector.fg_encoder.layer2.0.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.0/__module.kp_detector.fg_encoder.layer2.0.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.21 : Tensor = aten::batch_norm(%input.155, %self.kp_detector.fg_encoder.layer2.0.bn2.weight, %self.kp_detector.fg_encoder.layer2.0.bn2.bias, %self.kp_detector.fg_encoder.layer2.0.bn2.running_mean, %self.kp_detector.fg_encoder.layer2.0.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.0/__module.kp_detector.fg_encoder.layer2.0.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 (CaskConvolution)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - CaskConvolution has no valid tactics for this config, skipping
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %input.155 : Tensor = aten::_convolution(%234, %self.kp_detector.fg_encoder.layer2.0.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.0/__module.kp_detector.fg_encoder.layer2.0.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.21 : Tensor = aten::batch_norm(%input.155, %self.kp_detector.fg_encoder.layer2.0.bn2.weight, %self.kp_detector.fg_encoder.layer2.0.bn2.bias, %self.kp_detector.fg_encoder.layer2.0.bn2.running_mean, %self.kp_detector.fg_encoder.layer2.0.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.0/__module.kp_detector.fg_encoder.layer2.0.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 (CaskFlattenConvolution)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - CaskFlattenConvolution has no valid tactics for this config, skipping
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Half(16384,1:8,512,16) -> Half(16384,1:8,512,16) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - =============== Computing costs for 
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Float(262144,4096,64,1), Float(131072,1024,32,1) -> Float(131072,1024,32,1) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Float(262144,1,4096,64), Float(131072,1,4096,128) -> Float(131072,1,4096,128) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Float(65536,1:4,1024,16), Float(32768,1:4,1024,32) -> Float(32768,1:4,1024,32) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Half(262144,4096,64,1), Half(131072,1024,32,1) -> Half(131072,1024,32,1) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Half(131072,4096:2,64,1), Half(65536,1024:2,32,1) -> Half(65536,1024:2,32,1) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Half(65536,1:4,1024,16), Half(32768,1:4,1024,32) -> Half(32768,1:4,1024,32) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %input.271 : Tensor = aten::_convolution(%310, %self.kp_detector.fg_encoder.layer2.0.downsample.0.weight, %5, %25, %28, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.0/__module.kp_detector.fg_encoder.layer2.0.downsample/__module.kp_detector.fg_encoder.layer2.0.downsample.0 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %identity.13 : Tensor = aten::batch_norm(%input.271, %self.kp_detector.fg_encoder.layer2.0.downsample.1.weight, %self.kp_detector.fg_encoder.layer2.0.bn2.bias, %self.kp_detector.fg_encoder.layer2.0.downsample.1.running_mean, %self.kp_detector.fg_encoder.layer2.0.downsample.1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.0/__module.kp_detector.fg_encoder.layer2.0.downsample/__module.kp_detector.fg_encoder.layer2.0.downsample.1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %318 : Tensor = aten::add(%out.37, %identity.13, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.0 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %319 : Tensor = aten::relu(%318), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.0/__module.kp_detector.fg_encoder.layer2.0.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 (CaskConvolution)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - CaskConvolution has no valid tactics for this config, skipping
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %input.271 : Tensor = aten::_convolution(%310, %self.kp_detector.fg_encoder.layer2.0.downsample.0.weight, %5, %25, %28, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.0/__module.kp_detector.fg_encoder.layer2.0.downsample/__module.kp_detector.fg_encoder.layer2.0.downsample.0 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %identity.13 : Tensor = aten::batch_norm(%input.271, %self.kp_detector.fg_encoder.layer2.0.downsample.1.weight, %self.kp_detector.fg_encoder.layer2.0.bn2.bias, %self.kp_detector.fg_encoder.layer2.0.downsample.1.running_mean, %self.kp_detector.fg_encoder.layer2.0.downsample.1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.0/__module.kp_detector.fg_encoder.layer2.0.downsample/__module.kp_detector.fg_encoder.layer2.0.downsample.1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %318 : Tensor = aten::add(%out.37, %identity.13, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.0 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %319 : Tensor = aten::relu(%318), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.0/__module.kp_detector.fg_encoder.layer2.0.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 (CaskFlattenConvolution)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - CaskFlattenConvolution has no valid tactics for this config, skipping
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Half(32768,1:8,512,8), Float(131072,1024,32,1) -> Float(131072,1024,32,1) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %input.271 : Tensor = aten::_convolution(%310, %self.kp_detector.fg_encoder.layer2.0.downsample.0.weight, %5, %25, %28, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.0/__module.kp_detector.fg_encoder.layer2.0.downsample/__module.kp_detector.fg_encoder.layer2.0.downsample.0 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %identity.13 : Tensor = aten::batch_norm(%input.271, %self.kp_detector.fg_encoder.layer2.0.downsample.1.weight, %self.kp_detector.fg_encoder.layer2.0.bn2.bias, %self.kp_detector.fg_encoder.layer2.0.downsample.1.running_mean, %self.kp_detector.fg_encoder.layer2.0.downsample.1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.0/__module.kp_detector.fg_encoder.layer2.0.downsample/__module.kp_detector.fg_encoder.layer2.0.downsample.1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %318 : Tensor = aten::add(%out.37, %identity.13, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.0 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %319 : Tensor = aten::relu(%318), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.0/__module.kp_detector.fg_encoder.layer2.0.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 (CaskConvolution)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - CaskConvolution has no valid tactics for this config, skipping
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %input.271 : Tensor = aten::_convolution(%310, %self.kp_detector.fg_encoder.layer2.0.downsample.0.weight, %5, %25, %28, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.0/__module.kp_detector.fg_encoder.layer2.0.downsample/__module.kp_detector.fg_encoder.layer2.0.downsample.0 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %identity.13 : Tensor = aten::batch_norm(%input.271, %self.kp_detector.fg_encoder.layer2.0.downsample.1.weight, %self.kp_detector.fg_encoder.layer2.0.bn2.bias, %self.kp_detector.fg_encoder.layer2.0.downsample.1.running_mean, %self.kp_detector.fg_encoder.layer2.0.downsample.1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.0/__module.kp_detector.fg_encoder.layer2.0.downsample/__module.kp_detector.fg_encoder.layer2.0.downsample.1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %318 : Tensor = aten::add(%out.37, %identity.13, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.0 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %319 : Tensor = aten::relu(%318), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.0/__module.kp_detector.fg_encoder.layer2.0.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 (CaskFlattenConvolution)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - CaskFlattenConvolution has no valid tactics for this config, skipping
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Half(32768,1:8,512,8), Half(16384,1:8,512,16) -> Half(16384,1:8,512,16) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - =============== Computing costs for 
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Float(262144,4096,64,1), Float(131072,1024,32,1) -> Float(131072,1024,32,1) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Float(262144,1,4096,64), Float(131072,1,4096,128) -> Float(131072,1,4096,128) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Float(65536,1:4,1024,16), Float(32768,1:4,1024,32) -> Float(32768,1:4,1024,32) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Half(262144,4096,64,1), Half(131072,1024,32,1) -> Half(131072,1024,32,1) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Half(131072,4096:2,64,1), Half(65536,1024:2,32,1) -> Half(65536,1024:2,32,1) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Half(65536,1:4,1024,16), Half(32768,1:4,1024,32) -> Half(32768,1:4,1024,32) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %input.157 : Tensor = aten::_convolution(%231, %self.kp_detector.fg_encoder.layer2.0.downsample.0.weight, %5, %25, %28, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.0/__module.kp_detector.fg_encoder.layer2.0.downsample/__module.kp_detector.fg_encoder.layer2.0.downsample.0 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %identity.7 : Tensor = aten::batch_norm(%input.157, %self.kp_detector.fg_encoder.layer2.0.downsample.1.weight, %self.kp_detector.fg_encoder.layer2.0.bn2.bias, %self.kp_detector.fg_encoder.layer2.0.downsample.1.running_mean, %self.kp_detector.fg_encoder.layer2.0.downsample.1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.0/__module.kp_detector.fg_encoder.layer2.0.downsample/__module.kp_detector.fg_encoder.layer2.0.downsample.1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %239 : Tensor = aten::add(%out.21, %identity.7, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.0 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %240 : Tensor = aten::relu(%239), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.0/__module.kp_detector.fg_encoder.layer2.0.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 (CaskConvolution)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - CaskConvolution has no valid tactics for this config, skipping
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %input.157 : Tensor = aten::_convolution(%231, %self.kp_detector.fg_encoder.layer2.0.downsample.0.weight, %5, %25, %28, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.0/__module.kp_detector.fg_encoder.layer2.0.downsample/__module.kp_detector.fg_encoder.layer2.0.downsample.0 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %identity.7 : Tensor = aten::batch_norm(%input.157, %self.kp_detector.fg_encoder.layer2.0.downsample.1.weight, %self.kp_detector.fg_encoder.layer2.0.bn2.bias, %self.kp_detector.fg_encoder.layer2.0.downsample.1.running_mean, %self.kp_detector.fg_encoder.layer2.0.downsample.1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.0/__module.kp_detector.fg_encoder.layer2.0.downsample/__module.kp_detector.fg_encoder.layer2.0.downsample.1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %239 : Tensor = aten::add(%out.21, %identity.7, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.0 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %240 : Tensor = aten::relu(%239), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.0/__module.kp_detector.fg_encoder.layer2.0.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 (CaskFlattenConvolution)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - CaskFlattenConvolution has no valid tactics for this config, skipping
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Half(32768,1:8,512,8), Float(131072,1024,32,1) -> Float(131072,1024,32,1) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %input.157 : Tensor = aten::_convolution(%231, %self.kp_detector.fg_encoder.layer2.0.downsample.0.weight, %5, %25, %28, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.0/__module.kp_detector.fg_encoder.layer2.0.downsample/__module.kp_detector.fg_encoder.layer2.0.downsample.0 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %identity.7 : Tensor = aten::batch_norm(%input.157, %self.kp_detector.fg_encoder.layer2.0.downsample.1.weight, %self.kp_detector.fg_encoder.layer2.0.bn2.bias, %self.kp_detector.fg_encoder.layer2.0.downsample.1.running_mean, %self.kp_detector.fg_encoder.layer2.0.downsample.1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.0/__module.kp_detector.fg_encoder.layer2.0.downsample/__module.kp_detector.fg_encoder.layer2.0.downsample.1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %239 : Tensor = aten::add(%out.21, %identity.7, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.0 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %240 : Tensor = aten::relu(%239), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.0/__module.kp_detector.fg_encoder.layer2.0.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 (CaskConvolution)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - CaskConvolution has no valid tactics for this config, skipping
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %input.157 : Tensor = aten::_convolution(%231, %self.kp_detector.fg_encoder.layer2.0.downsample.0.weight, %5, %25, %28, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.0/__module.kp_detector.fg_encoder.layer2.0.downsample/__module.kp_detector.fg_encoder.layer2.0.downsample.0 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %identity.7 : Tensor = aten::batch_norm(%input.157, %self.kp_detector.fg_encoder.layer2.0.downsample.1.weight, %self.kp_detector.fg_encoder.layer2.0.bn2.bias, %self.kp_detector.fg_encoder.layer2.0.downsample.1.running_mean, %self.kp_detector.fg_encoder.layer2.0.downsample.1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.0/__module.kp_detector.fg_encoder.layer2.0.downsample/__module.kp_detector.fg_encoder.layer2.0.downsample.1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %239 : Tensor = aten::add(%out.21, %identity.7, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.0 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %240 : Tensor = aten::relu(%239), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.0/__module.kp_detector.fg_encoder.layer2.0.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 (CaskFlattenConvolution)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - CaskFlattenConvolution has no valid tactics for this config, skipping
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Half(32768,1:8,512,8), Half(16384,1:8,512,16) -> Half(16384,1:8,512,16) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - =============== Computing costs for 
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Float(131072,1024,32,1) -> Float(131072,1024,32,1) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Float(131072,1,4096,128) -> Float(131072,1,4096,128) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Float(32768,1:4,1024,32) -> Float(32768,1:4,1024,32) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Half(131072,1024,32,1) -> Half(131072,1024,32,1) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Half(65536,1024:2,32,1) -> Half(65536,1024:2,32,1) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %input.163 : Tensor = aten::_convolution(%240, %self.kp_detector.fg_encoder.layer2.1.conv1.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.1/__module.kp_detector.fg_encoder.layer2.1.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.165 : Tensor = aten::batch_norm(%input.163, %self.kp_detector.fg_encoder.layer2.1.bn1.weight, %self.kp_detector.fg_encoder.layer2.1.bn1.bias, %self.kp_detector.fg_encoder.layer2.1.bn1.running_mean, %self.kp_detector.fg_encoder.layer2.1.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.1/__module.kp_detector.fg_encoder.layer2.1.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %243 : Tensor = aten::relu(%input.165), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.1/__module.kp_detector.fg_encoder.layer2.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 (CaskConvolution)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - CaskConvolution has no valid tactics for this config, skipping
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %input.163 : Tensor = aten::_convolution(%240, %self.kp_detector.fg_encoder.layer2.1.conv1.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.1/__module.kp_detector.fg_encoder.layer2.1.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.165 : Tensor = aten::batch_norm(%input.163, %self.kp_detector.fg_encoder.layer2.1.bn1.weight, %self.kp_detector.fg_encoder.layer2.1.bn1.bias, %self.kp_detector.fg_encoder.layer2.1.bn1.running_mean, %self.kp_detector.fg_encoder.layer2.1.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.1/__module.kp_detector.fg_encoder.layer2.1.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %243 : Tensor = aten::relu(%input.165), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.1/__module.kp_detector.fg_encoder.layer2.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 (CaskFlattenConvolution)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - CaskFlattenConvolution has no valid tactics for this config, skipping
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Half(32768,1:4,1024,32) -> Half(32768,1:4,1024,32) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %input.163 : Tensor = aten::_convolution(%240, %self.kp_detector.fg_encoder.layer2.1.conv1.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.1/__module.kp_detector.fg_encoder.layer2.1.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.165 : Tensor = aten::batch_norm(%input.163, %self.kp_detector.fg_encoder.layer2.1.bn1.weight, %self.kp_detector.fg_encoder.layer2.1.bn1.bias, %self.kp_detector.fg_encoder.layer2.1.bn1.running_mean, %self.kp_detector.fg_encoder.layer2.1.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.1/__module.kp_detector.fg_encoder.layer2.1.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %243 : Tensor = aten::relu(%input.165), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.1/__module.kp_detector.fg_encoder.layer2.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 (CaskConvolution)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - CaskConvolution has no valid tactics for this config, skipping
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %input.163 : Tensor = aten::_convolution(%240, %self.kp_detector.fg_encoder.layer2.1.conv1.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.1/__module.kp_detector.fg_encoder.layer2.1.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.165 : Tensor = aten::batch_norm(%input.163, %self.kp_detector.fg_encoder.layer2.1.bn1.weight, %self.kp_detector.fg_encoder.layer2.1.bn1.bias, %self.kp_detector.fg_encoder.layer2.1.bn1.running_mean, %self.kp_detector.fg_encoder.layer2.1.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.1/__module.kp_detector.fg_encoder.layer2.1.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %243 : Tensor = aten::relu(%input.165), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.1/__module.kp_detector.fg_encoder.layer2.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 (CaskFlattenConvolution)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - CaskFlattenConvolution has no valid tactics for this config, skipping
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Half(16384,1:8,512,16) -> Float(131072,1024,32,1) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %input.163 : Tensor = aten::_convolution(%240, %self.kp_detector.fg_encoder.layer2.1.conv1.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.1/__module.kp_detector.fg_encoder.layer2.1.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.165 : Tensor = aten::batch_norm(%input.163, %self.kp_detector.fg_encoder.layer2.1.bn1.weight, %self.kp_detector.fg_encoder.layer2.1.bn1.bias, %self.kp_detector.fg_encoder.layer2.1.bn1.running_mean, %self.kp_detector.fg_encoder.layer2.1.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.1/__module.kp_detector.fg_encoder.layer2.1.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %243 : Tensor = aten::relu(%input.165), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.1/__module.kp_detector.fg_encoder.layer2.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 (CaskConvolution)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - CaskConvolution has no valid tactics for this config, skipping
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %input.163 : Tensor = aten::_convolution(%240, %self.kp_detector.fg_encoder.layer2.1.conv1.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.1/__module.kp_detector.fg_encoder.layer2.1.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.165 : Tensor = aten::batch_norm(%input.163, %self.kp_detector.fg_encoder.layer2.1.bn1.weight, %self.kp_detector.fg_encoder.layer2.1.bn1.bias, %self.kp_detector.fg_encoder.layer2.1.bn1.running_mean, %self.kp_detector.fg_encoder.layer2.1.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.1/__module.kp_detector.fg_encoder.layer2.1.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %243 : Tensor = aten::relu(%input.165), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.1/__module.kp_detector.fg_encoder.layer2.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 (CaskFlattenConvolution)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - CaskFlattenConvolution has no valid tactics for this config, skipping
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Half(16384,1:8,512,16) -> Half(16384,1:8,512,16) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - =============== Computing costs for 
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Float(131072,1024,32,1) -> Float(131072,1024,32,1) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Float(131072,1,4096,128) -> Float(131072,1,4096,128) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Float(32768,1:4,1024,32) -> Float(32768,1:4,1024,32) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Half(131072,1024,32,1) -> Half(131072,1024,32,1) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Half(65536,1024:2,32,1) -> Half(65536,1024:2,32,1) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %input.277 : Tensor = aten::_convolution(%319, %self.kp_detector.fg_encoder.layer2.1.conv1.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.1/__module.kp_detector.fg_encoder.layer2.1.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.279 : Tensor = aten::batch_norm(%input.277, %self.kp_detector.fg_encoder.layer2.1.bn1.weight, %self.kp_detector.fg_encoder.layer2.1.bn1.bias, %self.kp_detector.fg_encoder.layer2.1.bn1.running_mean, %self.kp_detector.fg_encoder.layer2.1.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.1/__module.kp_detector.fg_encoder.layer2.1.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %322 : Tensor = aten::relu(%input.279), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.1/__module.kp_detector.fg_encoder.layer2.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 (CaskConvolution)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - CaskConvolution has no valid tactics for this config, skipping
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %input.277 : Tensor = aten::_convolution(%319, %self.kp_detector.fg_encoder.layer2.1.conv1.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.1/__module.kp_detector.fg_encoder.layer2.1.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.279 : Tensor = aten::batch_norm(%input.277, %self.kp_detector.fg_encoder.layer2.1.bn1.weight, %self.kp_detector.fg_encoder.layer2.1.bn1.bias, %self.kp_detector.fg_encoder.layer2.1.bn1.running_mean, %self.kp_detector.fg_encoder.layer2.1.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.1/__module.kp_detector.fg_encoder.layer2.1.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %322 : Tensor = aten::relu(%input.279), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.1/__module.kp_detector.fg_encoder.layer2.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 (CaskFlattenConvolution)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - CaskFlattenConvolution has no valid tactics for this config, skipping
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Half(32768,1:4,1024,32) -> Half(32768,1:4,1024,32) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %input.277 : Tensor = aten::_convolution(%319, %self.kp_detector.fg_encoder.layer2.1.conv1.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.1/__module.kp_detector.fg_encoder.layer2.1.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.279 : Tensor = aten::batch_norm(%input.277, %self.kp_detector.fg_encoder.layer2.1.bn1.weight, %self.kp_detector.fg_encoder.layer2.1.bn1.bias, %self.kp_detector.fg_encoder.layer2.1.bn1.running_mean, %self.kp_detector.fg_encoder.layer2.1.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.1/__module.kp_detector.fg_encoder.layer2.1.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %322 : Tensor = aten::relu(%input.279), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.1/__module.kp_detector.fg_encoder.layer2.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 (CaskConvolution)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - CaskConvolution has no valid tactics for this config, skipping
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %input.277 : Tensor = aten::_convolution(%319, %self.kp_detector.fg_encoder.layer2.1.conv1.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.1/__module.kp_detector.fg_encoder.layer2.1.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.279 : Tensor = aten::batch_norm(%input.277, %self.kp_detector.fg_encoder.layer2.1.bn1.weight, %self.kp_detector.fg_encoder.layer2.1.bn1.bias, %self.kp_detector.fg_encoder.layer2.1.bn1.running_mean, %self.kp_detector.fg_encoder.layer2.1.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.1/__module.kp_detector.fg_encoder.layer2.1.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %322 : Tensor = aten::relu(%input.279), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.1/__module.kp_detector.fg_encoder.layer2.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 (CaskFlattenConvolution)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - CaskFlattenConvolution has no valid tactics for this config, skipping
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Half(16384,1:8,512,16) -> Float(131072,1024,32,1) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %input.277 : Tensor = aten::_convolution(%319, %self.kp_detector.fg_encoder.layer2.1.conv1.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.1/__module.kp_detector.fg_encoder.layer2.1.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.279 : Tensor = aten::batch_norm(%input.277, %self.kp_detector.fg_encoder.layer2.1.bn1.weight, %self.kp_detector.fg_encoder.layer2.1.bn1.bias, %self.kp_detector.fg_encoder.layer2.1.bn1.running_mean, %self.kp_detector.fg_encoder.layer2.1.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.1/__module.kp_detector.fg_encoder.layer2.1.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %322 : Tensor = aten::relu(%input.279), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.1/__module.kp_detector.fg_encoder.layer2.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 (CaskConvolution)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - CaskConvolution has no valid tactics for this config, skipping
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %input.277 : Tensor = aten::_convolution(%319, %self.kp_detector.fg_encoder.layer2.1.conv1.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.1/__module.kp_detector.fg_encoder.layer2.1.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.279 : Tensor = aten::batch_norm(%input.277, %self.kp_detector.fg_encoder.layer2.1.bn1.weight, %self.kp_detector.fg_encoder.layer2.1.bn1.bias, %self.kp_detector.fg_encoder.layer2.1.bn1.running_mean, %self.kp_detector.fg_encoder.layer2.1.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.1/__module.kp_detector.fg_encoder.layer2.1.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %322 : Tensor = aten::relu(%input.279), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.1/__module.kp_detector.fg_encoder.layer2.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 (CaskFlattenConvolution)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - CaskFlattenConvolution has no valid tactics for this config, skipping
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Half(16384,1:8,512,16) -> Half(16384,1:8,512,16) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - =============== Computing costs for 
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Float(131072,1024,32,1), Float(131072,1024,32,1) -> Float(131072,1024,32,1) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Float(131072,1,4096,128), Float(131072,1,4096,128) -> Float(131072,1,4096,128) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Float(32768,1:4,1024,32), Float(32768,1:4,1024,32) -> Float(32768,1:4,1024,32) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Half(131072,1024,32,1), Half(131072,1024,32,1) -> Half(131072,1024,32,1) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Half(65536,1024:2,32,1), Half(65536,1024:2,32,1) -> Half(65536,1024:2,32,1) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %input.283 : Tensor = aten::_convolution(%322, %self.kp_detector.fg_encoder.layer2.1.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.1/__module.kp_detector.fg_encoder.layer2.1.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.39 : Tensor = aten::batch_norm(%input.283, %self.kp_detector.fg_encoder.layer2.1.bn2.weight, %self.kp_detector.fg_encoder.layer2.1.bn2.bias, %self.kp_detector.fg_encoder.layer2.1.bn2.running_mean, %self.kp_detector.fg_encoder.layer2.1.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.1/__module.kp_detector.fg_encoder.layer2.1.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %325 : Tensor = aten::add(%out.39, %319, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.1 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %326 : Tensor = aten::relu(%325), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.1/__module.kp_detector.fg_encoder.layer2.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 (CaskConvolution)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - CaskConvolution has no valid tactics for this config, skipping
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %input.283 : Tensor = aten::_convolution(%322, %self.kp_detector.fg_encoder.layer2.1.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.1/__module.kp_detector.fg_encoder.layer2.1.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.39 : Tensor = aten::batch_norm(%input.283, %self.kp_detector.fg_encoder.layer2.1.bn2.weight, %self.kp_detector.fg_encoder.layer2.1.bn2.bias, %self.kp_detector.fg_encoder.layer2.1.bn2.running_mean, %self.kp_detector.fg_encoder.layer2.1.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.1/__module.kp_detector.fg_encoder.layer2.1.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %325 : Tensor = aten::add(%out.39, %319, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.1 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %326 : Tensor = aten::relu(%325), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.1/__module.kp_detector.fg_encoder.layer2.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 (CaskFlattenConvolution)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - CaskFlattenConvolution has no valid tactics for this config, skipping
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Half(32768,1:4,1024,32), Half(32768,1:4,1024,32) -> Half(32768,1:4,1024,32) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %input.283 : Tensor = aten::_convolution(%322, %self.kp_detector.fg_encoder.layer2.1.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.1/__module.kp_detector.fg_encoder.layer2.1.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.39 : Tensor = aten::batch_norm(%input.283, %self.kp_detector.fg_encoder.layer2.1.bn2.weight, %self.kp_detector.fg_encoder.layer2.1.bn2.bias, %self.kp_detector.fg_encoder.layer2.1.bn2.running_mean, %self.kp_detector.fg_encoder.layer2.1.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.1/__module.kp_detector.fg_encoder.layer2.1.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %325 : Tensor = aten::add(%out.39, %319, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.1 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %326 : Tensor = aten::relu(%325), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.1/__module.kp_detector.fg_encoder.layer2.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 (CaskConvolution)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - CaskConvolution has no valid tactics for this config, skipping
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %input.283 : Tensor = aten::_convolution(%322, %self.kp_detector.fg_encoder.layer2.1.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.1/__module.kp_detector.fg_encoder.layer2.1.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.39 : Tensor = aten::batch_norm(%input.283, %self.kp_detector.fg_encoder.layer2.1.bn2.weight, %self.kp_detector.fg_encoder.layer2.1.bn2.bias, %self.kp_detector.fg_encoder.layer2.1.bn2.running_mean, %self.kp_detector.fg_encoder.layer2.1.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.1/__module.kp_detector.fg_encoder.layer2.1.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %325 : Tensor = aten::add(%out.39, %319, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.1 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %326 : Tensor = aten::relu(%325), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.1/__module.kp_detector.fg_encoder.layer2.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 (CaskFlattenConvolution)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - CaskFlattenConvolution has no valid tactics for this config, skipping
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Half(16384,1:8,512,16), Float(131072,1024,32,1) -> Float(131072,1024,32,1) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %input.283 : Tensor = aten::_convolution(%322, %self.kp_detector.fg_encoder.layer2.1.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.1/__module.kp_detector.fg_encoder.layer2.1.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.39 : Tensor = aten::batch_norm(%input.283, %self.kp_detector.fg_encoder.layer2.1.bn2.weight, %self.kp_detector.fg_encoder.layer2.1.bn2.bias, %self.kp_detector.fg_encoder.layer2.1.bn2.running_mean, %self.kp_detector.fg_encoder.layer2.1.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.1/__module.kp_detector.fg_encoder.layer2.1.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %325 : Tensor = aten::add(%out.39, %319, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.1 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %326 : Tensor = aten::relu(%325), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.1/__module.kp_detector.fg_encoder.layer2.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 (CaskConvolution)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - CaskConvolution has no valid tactics for this config, skipping
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %input.283 : Tensor = aten::_convolution(%322, %self.kp_detector.fg_encoder.layer2.1.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.1/__module.kp_detector.fg_encoder.layer2.1.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.39 : Tensor = aten::batch_norm(%input.283, %self.kp_detector.fg_encoder.layer2.1.bn2.weight, %self.kp_detector.fg_encoder.layer2.1.bn2.bias, %self.kp_detector.fg_encoder.layer2.1.bn2.running_mean, %self.kp_detector.fg_encoder.layer2.1.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.1/__module.kp_detector.fg_encoder.layer2.1.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %325 : Tensor = aten::add(%out.39, %319, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.1 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %326 : Tensor = aten::relu(%325), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.1/__module.kp_detector.fg_encoder.layer2.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 (CaskFlattenConvolution)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - CaskFlattenConvolution has no valid tactics for this config, skipping
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Half(16384,1:8,512,16), Half(16384,1:8,512,16) -> Half(16384,1:8,512,16) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - =============== Computing costs for 
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Float(131072,1024,32,1), Float(131072,1024,32,1) -> Float(131072,1024,32,1) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Float(131072,1,4096,128), Float(131072,1,4096,128) -> Float(131072,1,4096,128) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Float(32768,1:4,1024,32), Float(32768,1:4,1024,32) -> Float(32768,1:4,1024,32) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Half(131072,1024,32,1), Half(131072,1024,32,1) -> Half(131072,1024,32,1) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Half(65536,1024:2,32,1), Half(65536,1024:2,32,1) -> Half(65536,1024:2,32,1) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %input.169 : Tensor = aten::_convolution(%243, %self.kp_detector.fg_encoder.layer2.1.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.1/__module.kp_detector.fg_encoder.layer2.1.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.23 : Tensor = aten::batch_norm(%input.169, %self.kp_detector.fg_encoder.layer2.1.bn2.weight, %self.kp_detector.fg_encoder.layer2.1.bn2.bias, %self.kp_detector.fg_encoder.layer2.1.bn2.running_mean, %self.kp_detector.fg_encoder.layer2.1.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.1/__module.kp_detector.fg_encoder.layer2.1.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %246 : Tensor = aten::add(%out.23, %240, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.1 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %247 : Tensor = aten::relu(%246), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.1/__module.kp_detector.fg_encoder.layer2.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 (CaskConvolution)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - CaskConvolution has no valid tactics for this config, skipping
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %input.169 : Tensor = aten::_convolution(%243, %self.kp_detector.fg_encoder.layer2.1.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.1/__module.kp_detector.fg_encoder.layer2.1.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.23 : Tensor = aten::batch_norm(%input.169, %self.kp_detector.fg_encoder.layer2.1.bn2.weight, %self.kp_detector.fg_encoder.layer2.1.bn2.bias, %self.kp_detector.fg_encoder.layer2.1.bn2.running_mean, %self.kp_detector.fg_encoder.layer2.1.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.1/__module.kp_detector.fg_encoder.layer2.1.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %246 : Tensor = aten::add(%out.23, %240, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.1 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %247 : Tensor = aten::relu(%246), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.1/__module.kp_detector.fg_encoder.layer2.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 (CaskFlattenConvolution)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - CaskFlattenConvolution has no valid tactics for this config, skipping
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Half(32768,1:4,1024,32), Half(32768,1:4,1024,32) -> Half(32768,1:4,1024,32) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %input.169 : Tensor = aten::_convolution(%243, %self.kp_detector.fg_encoder.layer2.1.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.1/__module.kp_detector.fg_encoder.layer2.1.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.23 : Tensor = aten::batch_norm(%input.169, %self.kp_detector.fg_encoder.layer2.1.bn2.weight, %self.kp_detector.fg_encoder.layer2.1.bn2.bias, %self.kp_detector.fg_encoder.layer2.1.bn2.running_mean, %self.kp_detector.fg_encoder.layer2.1.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.1/__module.kp_detector.fg_encoder.layer2.1.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %246 : Tensor = aten::add(%out.23, %240, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.1 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %247 : Tensor = aten::relu(%246), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.1/__module.kp_detector.fg_encoder.layer2.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 (CaskConvolution)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - CaskConvolution has no valid tactics for this config, skipping
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %input.169 : Tensor = aten::_convolution(%243, %self.kp_detector.fg_encoder.layer2.1.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.1/__module.kp_detector.fg_encoder.layer2.1.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.23 : Tensor = aten::batch_norm(%input.169, %self.kp_detector.fg_encoder.layer2.1.bn2.weight, %self.kp_detector.fg_encoder.layer2.1.bn2.bias, %self.kp_detector.fg_encoder.layer2.1.bn2.running_mean, %self.kp_detector.fg_encoder.layer2.1.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.1/__module.kp_detector.fg_encoder.layer2.1.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %246 : Tensor = aten::add(%out.23, %240, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.1 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %247 : Tensor = aten::relu(%246), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.1/__module.kp_detector.fg_encoder.layer2.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 (CaskFlattenConvolution)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - CaskFlattenConvolution has no valid tactics for this config, skipping
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Half(16384,1:8,512,16), Float(131072,1024,32,1) -> Float(131072,1024,32,1) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %input.169 : Tensor = aten::_convolution(%243, %self.kp_detector.fg_encoder.layer2.1.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.1/__module.kp_detector.fg_encoder.layer2.1.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.23 : Tensor = aten::batch_norm(%input.169, %self.kp_detector.fg_encoder.layer2.1.bn2.weight, %self.kp_detector.fg_encoder.layer2.1.bn2.bias, %self.kp_detector.fg_encoder.layer2.1.bn2.running_mean, %self.kp_detector.fg_encoder.layer2.1.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.1/__module.kp_detector.fg_encoder.layer2.1.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %246 : Tensor = aten::add(%out.23, %240, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.1 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %247 : Tensor = aten::relu(%246), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.1/__module.kp_detector.fg_encoder.layer2.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 (CaskConvolution)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - CaskConvolution has no valid tactics for this config, skipping
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %input.169 : Tensor = aten::_convolution(%243, %self.kp_detector.fg_encoder.layer2.1.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.1/__module.kp_detector.fg_encoder.layer2.1.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.23 : Tensor = aten::batch_norm(%input.169, %self.kp_detector.fg_encoder.layer2.1.bn2.weight, %self.kp_detector.fg_encoder.layer2.1.bn2.bias, %self.kp_detector.fg_encoder.layer2.1.bn2.running_mean, %self.kp_detector.fg_encoder.layer2.1.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.1/__module.kp_detector.fg_encoder.layer2.1.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %246 : Tensor = aten::add(%out.23, %240, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.1 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %247 : Tensor = aten::relu(%246), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.1/__module.kp_detector.fg_encoder.layer2.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 (CaskFlattenConvolution)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - CaskFlattenConvolution has no valid tactics for this config, skipping
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Half(16384,1:8,512,16), Half(16384,1:8,512,16) -> Half(16384,1:8,512,16) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - =============== Computing costs for 
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Float(131072,1024,32,1) -> Float(65536,256,16,1) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Float(131072,1,4096,128) -> Float(65536,1,4096,256) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Float(32768,1:4,1024,32) -> Float(16384,1:4,1024,64) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Half(131072,1024,32,1) -> Half(65536,256,16,1) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Half(65536,1024:2,32,1) -> Half(32768,256:2,16,1) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %input.175 : Tensor = aten::_convolution(%247, %self.kp_detector.fg_encoder.layer3.0.conv1.weight, %5, %25, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.177 : Tensor = aten::batch_norm(%input.175, %self.kp_detector.fg_encoder.layer3.0.bn1.weight, %self.kp_detector.fg_encoder.layer3.0.bn1.bias, %self.kp_detector.fg_encoder.layer3.0.bn1.running_mean, %self.kp_detector.fg_encoder.layer3.0.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %250 : Tensor = aten::relu(%input.177), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 (CaskConvolution)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - CaskConvolution has no valid tactics for this config, skipping
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %input.175 : Tensor = aten::_convolution(%247, %self.kp_detector.fg_encoder.layer3.0.conv1.weight, %5, %25, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.177 : Tensor = aten::batch_norm(%input.175, %self.kp_detector.fg_encoder.layer3.0.bn1.weight, %self.kp_detector.fg_encoder.layer3.0.bn1.bias, %self.kp_detector.fg_encoder.layer3.0.bn1.running_mean, %self.kp_detector.fg_encoder.layer3.0.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %250 : Tensor = aten::relu(%input.177), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 (CaskFlattenConvolution)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - CaskFlattenConvolution has no valid tactics for this config, skipping
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Half(32768,1:4,1024,32) -> Half(16384,1:4,1024,64) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %input.175 : Tensor = aten::_convolution(%247, %self.kp_detector.fg_encoder.layer3.0.conv1.weight, %5, %25, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.177 : Tensor = aten::batch_norm(%input.175, %self.kp_detector.fg_encoder.layer3.0.bn1.weight, %self.kp_detector.fg_encoder.layer3.0.bn1.bias, %self.kp_detector.fg_encoder.layer3.0.bn1.running_mean, %self.kp_detector.fg_encoder.layer3.0.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %250 : Tensor = aten::relu(%input.177), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 (CaskConvolution)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - CaskConvolution has no valid tactics for this config, skipping
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %input.175 : Tensor = aten::_convolution(%247, %self.kp_detector.fg_encoder.layer3.0.conv1.weight, %5, %25, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.177 : Tensor = aten::batch_norm(%input.175, %self.kp_detector.fg_encoder.layer3.0.bn1.weight, %self.kp_detector.fg_encoder.layer3.0.bn1.bias, %self.kp_detector.fg_encoder.layer3.0.bn1.running_mean, %self.kp_detector.fg_encoder.layer3.0.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %250 : Tensor = aten::relu(%input.177), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 (CaskFlattenConvolution)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - CaskFlattenConvolution has no valid tactics for this config, skipping
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Half(16384,1:8,512,16) -> Float(65536,256,16,1) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %input.175 : Tensor = aten::_convolution(%247, %self.kp_detector.fg_encoder.layer3.0.conv1.weight, %5, %25, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.177 : Tensor = aten::batch_norm(%input.175, %self.kp_detector.fg_encoder.layer3.0.bn1.weight, %self.kp_detector.fg_encoder.layer3.0.bn1.bias, %self.kp_detector.fg_encoder.layer3.0.bn1.running_mean, %self.kp_detector.fg_encoder.layer3.0.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %250 : Tensor = aten::relu(%input.177), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 (CaskConvolution)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - CaskConvolution has no valid tactics for this config, skipping
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %input.175 : Tensor = aten::_convolution(%247, %self.kp_detector.fg_encoder.layer3.0.conv1.weight, %5, %25, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.177 : Tensor = aten::batch_norm(%input.175, %self.kp_detector.fg_encoder.layer3.0.bn1.weight, %self.kp_detector.fg_encoder.layer3.0.bn1.bias, %self.kp_detector.fg_encoder.layer3.0.bn1.running_mean, %self.kp_detector.fg_encoder.layer3.0.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %250 : Tensor = aten::relu(%input.177), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 (CaskFlattenConvolution)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - CaskFlattenConvolution has no valid tactics for this config, skipping
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Half(16384,1:8,512,16) -> Half(8192,1:8,512,32) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - =============== Computing costs for 
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Float(131072,1024,32,1) -> Float(65536,256,16,1) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Float(131072,1,4096,128) -> Float(65536,1,4096,256) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Float(32768,1:4,1024,32) -> Float(16384,1:4,1024,64) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Half(131072,1024,32,1) -> Half(65536,256,16,1) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Half(65536,1024:2,32,1) -> Half(32768,256:2,16,1) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %input.289 : Tensor = aten::_convolution(%326, %self.kp_detector.fg_encoder.layer3.0.conv1.weight, %5, %25, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.291 : Tensor = aten::batch_norm(%input.289, %self.kp_detector.fg_encoder.layer3.0.bn1.weight, %self.kp_detector.fg_encoder.layer3.0.bn1.bias, %self.kp_detector.fg_encoder.layer3.0.bn1.running_mean, %self.kp_detector.fg_encoder.layer3.0.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %329 : Tensor = aten::relu(%input.291), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 (CaskConvolution)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - CaskConvolution has no valid tactics for this config, skipping
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %input.289 : Tensor = aten::_convolution(%326, %self.kp_detector.fg_encoder.layer3.0.conv1.weight, %5, %25, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.291 : Tensor = aten::batch_norm(%input.289, %self.kp_detector.fg_encoder.layer3.0.bn1.weight, %self.kp_detector.fg_encoder.layer3.0.bn1.bias, %self.kp_detector.fg_encoder.layer3.0.bn1.running_mean, %self.kp_detector.fg_encoder.layer3.0.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %329 : Tensor = aten::relu(%input.291), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 (CaskFlattenConvolution)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - CaskFlattenConvolution has no valid tactics for this config, skipping
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Half(32768,1:4,1024,32) -> Half(16384,1:4,1024,64) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %input.289 : Tensor = aten::_convolution(%326, %self.kp_detector.fg_encoder.layer3.0.conv1.weight, %5, %25, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.291 : Tensor = aten::batch_norm(%input.289, %self.kp_detector.fg_encoder.layer3.0.bn1.weight, %self.kp_detector.fg_encoder.layer3.0.bn1.bias, %self.kp_detector.fg_encoder.layer3.0.bn1.running_mean, %self.kp_detector.fg_encoder.layer3.0.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %329 : Tensor = aten::relu(%input.291), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 (CaskConvolution)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - CaskConvolution has no valid tactics for this config, skipping
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %input.289 : Tensor = aten::_convolution(%326, %self.kp_detector.fg_encoder.layer3.0.conv1.weight, %5, %25, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.291 : Tensor = aten::batch_norm(%input.289, %self.kp_detector.fg_encoder.layer3.0.bn1.weight, %self.kp_detector.fg_encoder.layer3.0.bn1.bias, %self.kp_detector.fg_encoder.layer3.0.bn1.running_mean, %self.kp_detector.fg_encoder.layer3.0.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %329 : Tensor = aten::relu(%input.291), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 (CaskFlattenConvolution)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - CaskFlattenConvolution has no valid tactics for this config, skipping
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Half(16384,1:8,512,16) -> Float(65536,256,16,1) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %input.289 : Tensor = aten::_convolution(%326, %self.kp_detector.fg_encoder.layer3.0.conv1.weight, %5, %25, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.291 : Tensor = aten::batch_norm(%input.289, %self.kp_detector.fg_encoder.layer3.0.bn1.weight, %self.kp_detector.fg_encoder.layer3.0.bn1.bias, %self.kp_detector.fg_encoder.layer3.0.bn1.running_mean, %self.kp_detector.fg_encoder.layer3.0.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %329 : Tensor = aten::relu(%input.291), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 (CaskConvolution)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - CaskConvolution has no valid tactics for this config, skipping
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %input.289 : Tensor = aten::_convolution(%326, %self.kp_detector.fg_encoder.layer3.0.conv1.weight, %5, %25, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.291 : Tensor = aten::batch_norm(%input.289, %self.kp_detector.fg_encoder.layer3.0.bn1.weight, %self.kp_detector.fg_encoder.layer3.0.bn1.bias, %self.kp_detector.fg_encoder.layer3.0.bn1.running_mean, %self.kp_detector.fg_encoder.layer3.0.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %329 : Tensor = aten::relu(%input.291), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 (CaskFlattenConvolution)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - CaskFlattenConvolution has no valid tactics for this config, skipping
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Half(16384,1:8,512,16) -> Half(8192,1:8,512,32) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - =============== Computing costs for 
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Float(65536,256,16,1) -> Float(65536,256,16,1) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Float(65536,1,4096,256) -> Float(65536,1,4096,256) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Float(16384,1:4,1024,64) -> Float(16384,1:4,1024,64) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Half(65536,256,16,1) -> Half(65536,256,16,1) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Half(32768,256:2,16,1) -> Half(32768,256:2,16,1) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %input.295 : Tensor = aten::_convolution(%329, %self.kp_detector.fg_encoder.layer3.0.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.41 : Tensor = aten::batch_norm(%input.295, %self.kp_detector.fg_encoder.layer3.0.bn2.weight, %self.kp_detector.fg_encoder.layer3.0.bn2.bias, %self.kp_detector.fg_encoder.layer3.0.bn2.running_mean, %self.kp_detector.fg_encoder.layer3.0.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 (CaskConvolution)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - CaskConvolution has no valid tactics for this config, skipping
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %input.295 : Tensor = aten::_convolution(%329, %self.kp_detector.fg_encoder.layer3.0.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.41 : Tensor = aten::batch_norm(%input.295, %self.kp_detector.fg_encoder.layer3.0.bn2.weight, %self.kp_detector.fg_encoder.layer3.0.bn2.bias, %self.kp_detector.fg_encoder.layer3.0.bn2.running_mean, %self.kp_detector.fg_encoder.layer3.0.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 (CaskFlattenConvolution)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - CaskFlattenConvolution has no valid tactics for this config, skipping
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Half(16384,1:4,1024,64) -> Half(16384,1:4,1024,64) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %input.295 : Tensor = aten::_convolution(%329, %self.kp_detector.fg_encoder.layer3.0.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.41 : Tensor = aten::batch_norm(%input.295, %self.kp_detector.fg_encoder.layer3.0.bn2.weight, %self.kp_detector.fg_encoder.layer3.0.bn2.bias, %self.kp_detector.fg_encoder.layer3.0.bn2.running_mean, %self.kp_detector.fg_encoder.layer3.0.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 (CaskConvolution)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - CaskConvolution has no valid tactics for this config, skipping
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %input.295 : Tensor = aten::_convolution(%329, %self.kp_detector.fg_encoder.layer3.0.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.41 : Tensor = aten::batch_norm(%input.295, %self.kp_detector.fg_encoder.layer3.0.bn2.weight, %self.kp_detector.fg_encoder.layer3.0.bn2.bias, %self.kp_detector.fg_encoder.layer3.0.bn2.running_mean, %self.kp_detector.fg_encoder.layer3.0.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 (CaskFlattenConvolution)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - CaskFlattenConvolution has no valid tactics for this config, skipping
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Half(8192,1:8,512,32) -> Float(65536,256,16,1) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %input.295 : Tensor = aten::_convolution(%329, %self.kp_detector.fg_encoder.layer3.0.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.41 : Tensor = aten::batch_norm(%input.295, %self.kp_detector.fg_encoder.layer3.0.bn2.weight, %self.kp_detector.fg_encoder.layer3.0.bn2.bias, %self.kp_detector.fg_encoder.layer3.0.bn2.running_mean, %self.kp_detector.fg_encoder.layer3.0.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 (CaskConvolution)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - CaskConvolution has no valid tactics for this config, skipping
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %input.295 : Tensor = aten::_convolution(%329, %self.kp_detector.fg_encoder.layer3.0.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.41 : Tensor = aten::batch_norm(%input.295, %self.kp_detector.fg_encoder.layer3.0.bn2.weight, %self.kp_detector.fg_encoder.layer3.0.bn2.bias, %self.kp_detector.fg_encoder.layer3.0.bn2.running_mean, %self.kp_detector.fg_encoder.layer3.0.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 (CaskFlattenConvolution)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - CaskFlattenConvolution has no valid tactics for this config, skipping
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Half(8192,1:8,512,32) -> Half(8192,1:8,512,32) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - =============== Computing costs for 
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Float(65536,256,16,1) -> Float(65536,256,16,1) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Float(65536,1,4096,256) -> Float(65536,1,4096,256) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Float(16384,1:4,1024,64) -> Float(16384,1:4,1024,64) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Half(65536,256,16,1) -> Half(65536,256,16,1) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Half(32768,256:2,16,1) -> Half(32768,256:2,16,1) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %input.181 : Tensor = aten::_convolution(%250, %self.kp_detector.fg_encoder.layer3.0.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.25 : Tensor = aten::batch_norm(%input.181, %self.kp_detector.fg_encoder.layer3.0.bn2.weight, %self.kp_detector.fg_encoder.layer3.0.bn2.bias, %self.kp_detector.fg_encoder.layer3.0.bn2.running_mean, %self.kp_detector.fg_encoder.layer3.0.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 (CaskConvolution)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - CaskConvolution has no valid tactics for this config, skipping
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %input.181 : Tensor = aten::_convolution(%250, %self.kp_detector.fg_encoder.layer3.0.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.25 : Tensor = aten::batch_norm(%input.181, %self.kp_detector.fg_encoder.layer3.0.bn2.weight, %self.kp_detector.fg_encoder.layer3.0.bn2.bias, %self.kp_detector.fg_encoder.layer3.0.bn2.running_mean, %self.kp_detector.fg_encoder.layer3.0.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 (CaskFlattenConvolution)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - CaskFlattenConvolution has no valid tactics for this config, skipping
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Half(16384,1:4,1024,64) -> Half(16384,1:4,1024,64) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %input.181 : Tensor = aten::_convolution(%250, %self.kp_detector.fg_encoder.layer3.0.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.25 : Tensor = aten::batch_norm(%input.181, %self.kp_detector.fg_encoder.layer3.0.bn2.weight, %self.kp_detector.fg_encoder.layer3.0.bn2.bias, %self.kp_detector.fg_encoder.layer3.0.bn2.running_mean, %self.kp_detector.fg_encoder.layer3.0.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 (CaskConvolution)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - CaskConvolution has no valid tactics for this config, skipping
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %input.181 : Tensor = aten::_convolution(%250, %self.kp_detector.fg_encoder.layer3.0.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.25 : Tensor = aten::batch_norm(%input.181, %self.kp_detector.fg_encoder.layer3.0.bn2.weight, %self.kp_detector.fg_encoder.layer3.0.bn2.bias, %self.kp_detector.fg_encoder.layer3.0.bn2.running_mean, %self.kp_detector.fg_encoder.layer3.0.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 (CaskFlattenConvolution)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - CaskFlattenConvolution has no valid tactics for this config, skipping
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Half(8192,1:8,512,32) -> Float(65536,256,16,1) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %input.181 : Tensor = aten::_convolution(%250, %self.kp_detector.fg_encoder.layer3.0.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.25 : Tensor = aten::batch_norm(%input.181, %self.kp_detector.fg_encoder.layer3.0.bn2.weight, %self.kp_detector.fg_encoder.layer3.0.bn2.bias, %self.kp_detector.fg_encoder.layer3.0.bn2.running_mean, %self.kp_detector.fg_encoder.layer3.0.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 (CaskConvolution)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - CaskConvolution has no valid tactics for this config, skipping
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %input.181 : Tensor = aten::_convolution(%250, %self.kp_detector.fg_encoder.layer3.0.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.25 : Tensor = aten::batch_norm(%input.181, %self.kp_detector.fg_encoder.layer3.0.bn2.weight, %self.kp_detector.fg_encoder.layer3.0.bn2.bias, %self.kp_detector.fg_encoder.layer3.0.bn2.running_mean, %self.kp_detector.fg_encoder.layer3.0.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 (CaskFlattenConvolution)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - CaskFlattenConvolution has no valid tactics for this config, skipping
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Half(8192,1:8,512,32) -> Half(8192,1:8,512,32) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - =============== Computing costs for 
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Float(131072,1024,32,1), Float(65536,256,16,1) -> Float(65536,256,16,1) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Float(131072,1,4096,128), Float(65536,1,4096,256) -> Float(65536,1,4096,256) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Float(32768,1:4,1024,32), Float(16384,1:4,1024,64) -> Float(16384,1:4,1024,64) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Half(131072,1024,32,1), Half(65536,256,16,1) -> Half(65536,256,16,1) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Half(65536,1024:2,32,1), Half(32768,256:2,16,1) -> Half(32768,256:2,16,1) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %input.297 : Tensor = aten::_convolution(%326, %self.kp_detector.fg_encoder.layer3.0.downsample.0.weight, %5, %25, %28, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.downsample/__module.kp_detector.fg_encoder.layer3.0.downsample.0 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %identity.15 : Tensor = aten::batch_norm(%input.297, %self.kp_detector.fg_encoder.layer3.0.downsample.1.weight, %self.kp_detector.fg_encoder.layer3.0.bn2.bias, %self.kp_detector.fg_encoder.layer3.0.downsample.1.running_mean, %self.kp_detector.fg_encoder.layer3.0.downsample.1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.downsample/__module.kp_detector.fg_encoder.layer3.0.downsample.1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %334 : Tensor = aten::add(%out.41, %identity.15, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %335 : Tensor = aten::relu(%334), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 (CaskConvolution)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - CaskConvolution has no valid tactics for this config, skipping
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %input.297 : Tensor = aten::_convolution(%326, %self.kp_detector.fg_encoder.layer3.0.downsample.0.weight, %5, %25, %28, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.downsample/__module.kp_detector.fg_encoder.layer3.0.downsample.0 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %identity.15 : Tensor = aten::batch_norm(%input.297, %self.kp_detector.fg_encoder.layer3.0.downsample.1.weight, %self.kp_detector.fg_encoder.layer3.0.bn2.bias, %self.kp_detector.fg_encoder.layer3.0.downsample.1.running_mean, %self.kp_detector.fg_encoder.layer3.0.downsample.1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.downsample/__module.kp_detector.fg_encoder.layer3.0.downsample.1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %334 : Tensor = aten::add(%out.41, %identity.15, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %335 : Tensor = aten::relu(%334), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 (CaskFlattenConvolution)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - CaskFlattenConvolution has no valid tactics for this config, skipping
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Half(32768,1:4,1024,32), Half(16384,1:4,1024,64) -> Half(16384,1:4,1024,64) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %input.297 : Tensor = aten::_convolution(%326, %self.kp_detector.fg_encoder.layer3.0.downsample.0.weight, %5, %25, %28, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.downsample/__module.kp_detector.fg_encoder.layer3.0.downsample.0 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %identity.15 : Tensor = aten::batch_norm(%input.297, %self.kp_detector.fg_encoder.layer3.0.downsample.1.weight, %self.kp_detector.fg_encoder.layer3.0.bn2.bias, %self.kp_detector.fg_encoder.layer3.0.downsample.1.running_mean, %self.kp_detector.fg_encoder.layer3.0.downsample.1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.downsample/__module.kp_detector.fg_encoder.layer3.0.downsample.1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %334 : Tensor = aten::add(%out.41, %identity.15, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %335 : Tensor = aten::relu(%334), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 (CaskConvolution)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - CaskConvolution has no valid tactics for this config, skipping
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %input.297 : Tensor = aten::_convolution(%326, %self.kp_detector.fg_encoder.layer3.0.downsample.0.weight, %5, %25, %28, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.downsample/__module.kp_detector.fg_encoder.layer3.0.downsample.0 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %identity.15 : Tensor = aten::batch_norm(%input.297, %self.kp_detector.fg_encoder.layer3.0.downsample.1.weight, %self.kp_detector.fg_encoder.layer3.0.bn2.bias, %self.kp_detector.fg_encoder.layer3.0.downsample.1.running_mean, %self.kp_detector.fg_encoder.layer3.0.downsample.1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.downsample/__module.kp_detector.fg_encoder.layer3.0.downsample.1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %334 : Tensor = aten::add(%out.41, %identity.15, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %335 : Tensor = aten::relu(%334), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 (CaskFlattenConvolution)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - CaskFlattenConvolution has no valid tactics for this config, skipping
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Half(16384,1:8,512,16), Float(65536,256,16,1) -> Float(65536,256,16,1) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %input.297 : Tensor = aten::_convolution(%326, %self.kp_detector.fg_encoder.layer3.0.downsample.0.weight, %5, %25, %28, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.downsample/__module.kp_detector.fg_encoder.layer3.0.downsample.0 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %identity.15 : Tensor = aten::batch_norm(%input.297, %self.kp_detector.fg_encoder.layer3.0.downsample.1.weight, %self.kp_detector.fg_encoder.layer3.0.bn2.bias, %self.kp_detector.fg_encoder.layer3.0.downsample.1.running_mean, %self.kp_detector.fg_encoder.layer3.0.downsample.1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.downsample/__module.kp_detector.fg_encoder.layer3.0.downsample.1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %334 : Tensor = aten::add(%out.41, %identity.15, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %335 : Tensor = aten::relu(%334), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 (CaskConvolution)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - CaskConvolution has no valid tactics for this config, skipping
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %input.297 : Tensor = aten::_convolution(%326, %self.kp_detector.fg_encoder.layer3.0.downsample.0.weight, %5, %25, %28, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.downsample/__module.kp_detector.fg_encoder.layer3.0.downsample.0 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %identity.15 : Tensor = aten::batch_norm(%input.297, %self.kp_detector.fg_encoder.layer3.0.downsample.1.weight, %self.kp_detector.fg_encoder.layer3.0.bn2.bias, %self.kp_detector.fg_encoder.layer3.0.downsample.1.running_mean, %self.kp_detector.fg_encoder.layer3.0.downsample.1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.downsample/__module.kp_detector.fg_encoder.layer3.0.downsample.1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %334 : Tensor = aten::add(%out.41, %identity.15, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %335 : Tensor = aten::relu(%334), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 (CaskFlattenConvolution)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - CaskFlattenConvolution has no valid tactics for this config, skipping
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Half(16384,1:8,512,16), Half(8192,1:8,512,32) -> Half(8192,1:8,512,32) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - =============== Computing costs for 
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Float(131072,1024,32,1), Float(65536,256,16,1) -> Float(65536,256,16,1) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Float(131072,1,4096,128), Float(65536,1,4096,256) -> Float(65536,1,4096,256) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Float(32768,1:4,1024,32), Float(16384,1:4,1024,64) -> Float(16384,1:4,1024,64) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Half(131072,1024,32,1), Half(65536,256,16,1) -> Half(65536,256,16,1) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Half(65536,1024:2,32,1), Half(32768,256:2,16,1) -> Half(32768,256:2,16,1) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %input.183 : Tensor = aten::_convolution(%247, %self.kp_detector.fg_encoder.layer3.0.downsample.0.weight, %5, %25, %28, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.downsample/__module.kp_detector.fg_encoder.layer3.0.downsample.0 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %identity.9 : Tensor = aten::batch_norm(%input.183, %self.kp_detector.fg_encoder.layer3.0.downsample.1.weight, %self.kp_detector.fg_encoder.layer3.0.bn2.bias, %self.kp_detector.fg_encoder.layer3.0.downsample.1.running_mean, %self.kp_detector.fg_encoder.layer3.0.downsample.1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.downsample/__module.kp_detector.fg_encoder.layer3.0.downsample.1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %255 : Tensor = aten::add(%out.25, %identity.9, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %256 : Tensor = aten::relu(%255), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 (CaskConvolution)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - CaskConvolution has no valid tactics for this config, skipping
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %input.183 : Tensor = aten::_convolution(%247, %self.kp_detector.fg_encoder.layer3.0.downsample.0.weight, %5, %25, %28, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.downsample/__module.kp_detector.fg_encoder.layer3.0.downsample.0 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %identity.9 : Tensor = aten::batch_norm(%input.183, %self.kp_detector.fg_encoder.layer3.0.downsample.1.weight, %self.kp_detector.fg_encoder.layer3.0.bn2.bias, %self.kp_detector.fg_encoder.layer3.0.downsample.1.running_mean, %self.kp_detector.fg_encoder.layer3.0.downsample.1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.downsample/__module.kp_detector.fg_encoder.layer3.0.downsample.1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %255 : Tensor = aten::add(%out.25, %identity.9, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %256 : Tensor = aten::relu(%255), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 (CaskFlattenConvolution)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - CaskFlattenConvolution has no valid tactics for this config, skipping
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Half(32768,1:4,1024,32), Half(16384,1:4,1024,64) -> Half(16384,1:4,1024,64) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %input.183 : Tensor = aten::_convolution(%247, %self.kp_detector.fg_encoder.layer3.0.downsample.0.weight, %5, %25, %28, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.downsample/__module.kp_detector.fg_encoder.layer3.0.downsample.0 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %identity.9 : Tensor = aten::batch_norm(%input.183, %self.kp_detector.fg_encoder.layer3.0.downsample.1.weight, %self.kp_detector.fg_encoder.layer3.0.bn2.bias, %self.kp_detector.fg_encoder.layer3.0.downsample.1.running_mean, %self.kp_detector.fg_encoder.layer3.0.downsample.1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.downsample/__module.kp_detector.fg_encoder.layer3.0.downsample.1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %255 : Tensor = aten::add(%out.25, %identity.9, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %256 : Tensor = aten::relu(%255), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 (CaskConvolution)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - CaskConvolution has no valid tactics for this config, skipping
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %input.183 : Tensor = aten::_convolution(%247, %self.kp_detector.fg_encoder.layer3.0.downsample.0.weight, %5, %25, %28, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.downsample/__module.kp_detector.fg_encoder.layer3.0.downsample.0 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %identity.9 : Tensor = aten::batch_norm(%input.183, %self.kp_detector.fg_encoder.layer3.0.downsample.1.weight, %self.kp_detector.fg_encoder.layer3.0.bn2.bias, %self.kp_detector.fg_encoder.layer3.0.downsample.1.running_mean, %self.kp_detector.fg_encoder.layer3.0.downsample.1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.downsample/__module.kp_detector.fg_encoder.layer3.0.downsample.1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %255 : Tensor = aten::add(%out.25, %identity.9, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %256 : Tensor = aten::relu(%255), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 (CaskFlattenConvolution)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - CaskFlattenConvolution has no valid tactics for this config, skipping
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Half(16384,1:8,512,16), Float(65536,256,16,1) -> Float(65536,256,16,1) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %input.183 : Tensor = aten::_convolution(%247, %self.kp_detector.fg_encoder.layer3.0.downsample.0.weight, %5, %25, %28, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.downsample/__module.kp_detector.fg_encoder.layer3.0.downsample.0 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %identity.9 : Tensor = aten::batch_norm(%input.183, %self.kp_detector.fg_encoder.layer3.0.downsample.1.weight, %self.kp_detector.fg_encoder.layer3.0.bn2.bias, %self.kp_detector.fg_encoder.layer3.0.downsample.1.running_mean, %self.kp_detector.fg_encoder.layer3.0.downsample.1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.downsample/__module.kp_detector.fg_encoder.layer3.0.downsample.1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %255 : Tensor = aten::add(%out.25, %identity.9, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %256 : Tensor = aten::relu(%255), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 (CaskConvolution)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - CaskConvolution has no valid tactics for this config, skipping
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %input.183 : Tensor = aten::_convolution(%247, %self.kp_detector.fg_encoder.layer3.0.downsample.0.weight, %5, %25, %28, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.downsample/__module.kp_detector.fg_encoder.layer3.0.downsample.0 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %identity.9 : Tensor = aten::batch_norm(%input.183, %self.kp_detector.fg_encoder.layer3.0.downsample.1.weight, %self.kp_detector.fg_encoder.layer3.0.bn2.bias, %self.kp_detector.fg_encoder.layer3.0.downsample.1.running_mean, %self.kp_detector.fg_encoder.layer3.0.downsample.1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.downsample/__module.kp_detector.fg_encoder.layer3.0.downsample.1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %255 : Tensor = aten::add(%out.25, %identity.9, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %256 : Tensor = aten::relu(%255), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 (CaskFlattenConvolution)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - CaskFlattenConvolution has no valid tactics for this config, skipping
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Half(16384,1:8,512,16), Half(8192,1:8,512,32) -> Half(8192,1:8,512,32) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - =============== Computing costs for 
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Float(65536,256,16,1) -> Float(65536,256,16,1) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Float(65536,1,4096,256) -> Float(65536,1,4096,256) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Float(16384,1:4,1024,64) -> Float(16384,1:4,1024,64) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Half(65536,256,16,1) -> Half(65536,256,16,1) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Half(32768,256:2,16,1) -> Half(32768,256:2,16,1) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %input.189 : Tensor = aten::_convolution(%256, %self.kp_detector.fg_encoder.layer3.1.conv1.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1/__module.kp_detector.fg_encoder.layer3.1.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.191 : Tensor = aten::batch_norm(%input.189, %self.kp_detector.fg_encoder.layer3.1.bn1.weight, %self.kp_detector.fg_encoder.layer3.1.bn1.bias, %self.kp_detector.fg_encoder.layer3.1.bn1.running_mean, %self.kp_detector.fg_encoder.layer3.1.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1/__module.kp_detector.fg_encoder.layer3.1.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %259 : Tensor = aten::relu(%input.191), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1/__module.kp_detector.fg_encoder.layer3.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 (CaskConvolution)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - CaskConvolution has no valid tactics for this config, skipping
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %input.189 : Tensor = aten::_convolution(%256, %self.kp_detector.fg_encoder.layer3.1.conv1.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1/__module.kp_detector.fg_encoder.layer3.1.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.191 : Tensor = aten::batch_norm(%input.189, %self.kp_detector.fg_encoder.layer3.1.bn1.weight, %self.kp_detector.fg_encoder.layer3.1.bn1.bias, %self.kp_detector.fg_encoder.layer3.1.bn1.running_mean, %self.kp_detector.fg_encoder.layer3.1.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1/__module.kp_detector.fg_encoder.layer3.1.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %259 : Tensor = aten::relu(%input.191), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1/__module.kp_detector.fg_encoder.layer3.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 (CaskFlattenConvolution)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - CaskFlattenConvolution has no valid tactics for this config, skipping
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Half(16384,1:4,1024,64) -> Half(16384,1:4,1024,64) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %input.189 : Tensor = aten::_convolution(%256, %self.kp_detector.fg_encoder.layer3.1.conv1.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1/__module.kp_detector.fg_encoder.layer3.1.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.191 : Tensor = aten::batch_norm(%input.189, %self.kp_detector.fg_encoder.layer3.1.bn1.weight, %self.kp_detector.fg_encoder.layer3.1.bn1.bias, %self.kp_detector.fg_encoder.layer3.1.bn1.running_mean, %self.kp_detector.fg_encoder.layer3.1.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1/__module.kp_detector.fg_encoder.layer3.1.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %259 : Tensor = aten::relu(%input.191), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1/__module.kp_detector.fg_encoder.layer3.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 (CaskConvolution)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - CaskConvolution has no valid tactics for this config, skipping
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %input.189 : Tensor = aten::_convolution(%256, %self.kp_detector.fg_encoder.layer3.1.conv1.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1/__module.kp_detector.fg_encoder.layer3.1.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.191 : Tensor = aten::batch_norm(%input.189, %self.kp_detector.fg_encoder.layer3.1.bn1.weight, %self.kp_detector.fg_encoder.layer3.1.bn1.bias, %self.kp_detector.fg_encoder.layer3.1.bn1.running_mean, %self.kp_detector.fg_encoder.layer3.1.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1/__module.kp_detector.fg_encoder.layer3.1.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %259 : Tensor = aten::relu(%input.191), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1/__module.kp_detector.fg_encoder.layer3.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 (CaskFlattenConvolution)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - CaskFlattenConvolution has no valid tactics for this config, skipping
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Half(8192,1:8,512,32) -> Float(65536,256,16,1) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %input.189 : Tensor = aten::_convolution(%256, %self.kp_detector.fg_encoder.layer3.1.conv1.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1/__module.kp_detector.fg_encoder.layer3.1.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.191 : Tensor = aten::batch_norm(%input.189, %self.kp_detector.fg_encoder.layer3.1.bn1.weight, %self.kp_detector.fg_encoder.layer3.1.bn1.bias, %self.kp_detector.fg_encoder.layer3.1.bn1.running_mean, %self.kp_detector.fg_encoder.layer3.1.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1/__module.kp_detector.fg_encoder.layer3.1.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %259 : Tensor = aten::relu(%input.191), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1/__module.kp_detector.fg_encoder.layer3.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 (CaskConvolution)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - CaskConvolution has no valid tactics for this config, skipping
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %input.189 : Tensor = aten::_convolution(%256, %self.kp_detector.fg_encoder.layer3.1.conv1.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1/__module.kp_detector.fg_encoder.layer3.1.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.191 : Tensor = aten::batch_norm(%input.189, %self.kp_detector.fg_encoder.layer3.1.bn1.weight, %self.kp_detector.fg_encoder.layer3.1.bn1.bias, %self.kp_detector.fg_encoder.layer3.1.bn1.running_mean, %self.kp_detector.fg_encoder.layer3.1.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1/__module.kp_detector.fg_encoder.layer3.1.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %259 : Tensor = aten::relu(%input.191), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1/__module.kp_detector.fg_encoder.layer3.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 (CaskFlattenConvolution)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - CaskFlattenConvolution has no valid tactics for this config, skipping
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Half(8192,1:8,512,32) -> Half(8192,1:8,512,32) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - =============== Computing costs for 
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Float(65536,256,16,1) -> Float(65536,256,16,1) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Float(65536,1,4096,256) -> Float(65536,1,4096,256) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Float(16384,1:4,1024,64) -> Float(16384,1:4,1024,64) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Half(65536,256,16,1) -> Half(65536,256,16,1) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Half(32768,256:2,16,1) -> Half(32768,256:2,16,1) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %input.303 : Tensor = aten::_convolution(%335, %self.kp_detector.fg_encoder.layer3.1.conv1.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1/__module.kp_detector.fg_encoder.layer3.1.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.305 : Tensor = aten::batch_norm(%input.303, %self.kp_detector.fg_encoder.layer3.1.bn1.weight, %self.kp_detector.fg_encoder.layer3.1.bn1.bias, %self.kp_detector.fg_encoder.layer3.1.bn1.running_mean, %self.kp_detector.fg_encoder.layer3.1.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1/__module.kp_detector.fg_encoder.layer3.1.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %338 : Tensor = aten::relu(%input.305), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1/__module.kp_detector.fg_encoder.layer3.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 (CaskConvolution)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - CaskConvolution has no valid tactics for this config, skipping
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %input.303 : Tensor = aten::_convolution(%335, %self.kp_detector.fg_encoder.layer3.1.conv1.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1/__module.kp_detector.fg_encoder.layer3.1.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.305 : Tensor = aten::batch_norm(%input.303, %self.kp_detector.fg_encoder.layer3.1.bn1.weight, %self.kp_detector.fg_encoder.layer3.1.bn1.bias, %self.kp_detector.fg_encoder.layer3.1.bn1.running_mean, %self.kp_detector.fg_encoder.layer3.1.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1/__module.kp_detector.fg_encoder.layer3.1.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %338 : Tensor = aten::relu(%input.305), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1/__module.kp_detector.fg_encoder.layer3.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 (CaskFlattenConvolution)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - CaskFlattenConvolution has no valid tactics for this config, skipping
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Half(16384,1:4,1024,64) -> Half(16384,1:4,1024,64) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %input.303 : Tensor = aten::_convolution(%335, %self.kp_detector.fg_encoder.layer3.1.conv1.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1/__module.kp_detector.fg_encoder.layer3.1.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.305 : Tensor = aten::batch_norm(%input.303, %self.kp_detector.fg_encoder.layer3.1.bn1.weight, %self.kp_detector.fg_encoder.layer3.1.bn1.bias, %self.kp_detector.fg_encoder.layer3.1.bn1.running_mean, %self.kp_detector.fg_encoder.layer3.1.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1/__module.kp_detector.fg_encoder.layer3.1.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %338 : Tensor = aten::relu(%input.305), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1/__module.kp_detector.fg_encoder.layer3.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 (CaskConvolution)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - CaskConvolution has no valid tactics for this config, skipping
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %input.303 : Tensor = aten::_convolution(%335, %self.kp_detector.fg_encoder.layer3.1.conv1.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1/__module.kp_detector.fg_encoder.layer3.1.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.305 : Tensor = aten::batch_norm(%input.303, %self.kp_detector.fg_encoder.layer3.1.bn1.weight, %self.kp_detector.fg_encoder.layer3.1.bn1.bias, %self.kp_detector.fg_encoder.layer3.1.bn1.running_mean, %self.kp_detector.fg_encoder.layer3.1.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1/__module.kp_detector.fg_encoder.layer3.1.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %338 : Tensor = aten::relu(%input.305), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1/__module.kp_detector.fg_encoder.layer3.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 (CaskFlattenConvolution)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - CaskFlattenConvolution has no valid tactics for this config, skipping
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Half(8192,1:8,512,32) -> Float(65536,256,16,1) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %input.303 : Tensor = aten::_convolution(%335, %self.kp_detector.fg_encoder.layer3.1.conv1.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1/__module.kp_detector.fg_encoder.layer3.1.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.305 : Tensor = aten::batch_norm(%input.303, %self.kp_detector.fg_encoder.layer3.1.bn1.weight, %self.kp_detector.fg_encoder.layer3.1.bn1.bias, %self.kp_detector.fg_encoder.layer3.1.bn1.running_mean, %self.kp_detector.fg_encoder.layer3.1.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1/__module.kp_detector.fg_encoder.layer3.1.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %338 : Tensor = aten::relu(%input.305), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1/__module.kp_detector.fg_encoder.layer3.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 (CaskConvolution)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - CaskConvolution has no valid tactics for this config, skipping
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %input.303 : Tensor = aten::_convolution(%335, %self.kp_detector.fg_encoder.layer3.1.conv1.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1/__module.kp_detector.fg_encoder.layer3.1.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.305 : Tensor = aten::batch_norm(%input.303, %self.kp_detector.fg_encoder.layer3.1.bn1.weight, %self.kp_detector.fg_encoder.layer3.1.bn1.bias, %self.kp_detector.fg_encoder.layer3.1.bn1.running_mean, %self.kp_detector.fg_encoder.layer3.1.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1/__module.kp_detector.fg_encoder.layer3.1.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %338 : Tensor = aten::relu(%input.305), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1/__module.kp_detector.fg_encoder.layer3.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 (CaskFlattenConvolution)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - CaskFlattenConvolution has no valid tactics for this config, skipping
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Half(8192,1:8,512,32) -> Half(8192,1:8,512,32) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - =============== Computing costs for 
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Float(65536,256,16,1), Float(65536,256,16,1) -> Float(65536,256,16,1) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Float(65536,1,4096,256), Float(65536,1,4096,256) -> Float(65536,1,4096,256) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Float(16384,1:4,1024,64), Float(16384,1:4,1024,64) -> Float(16384,1:4,1024,64) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Half(65536,256,16,1), Half(65536,256,16,1) -> Half(65536,256,16,1) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Half(32768,256:2,16,1), Half(32768,256:2,16,1) -> Half(32768,256:2,16,1) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %input.309 : Tensor = aten::_convolution(%338, %self.kp_detector.fg_encoder.layer3.1.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1/__module.kp_detector.fg_encoder.layer3.1.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.43 : Tensor = aten::batch_norm(%input.309, %self.kp_detector.fg_encoder.layer3.1.bn2.weight, %self.kp_detector.fg_encoder.layer3.1.bn2.bias, %self.kp_detector.fg_encoder.layer3.1.bn2.running_mean, %self.kp_detector.fg_encoder.layer3.1.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1/__module.kp_detector.fg_encoder.layer3.1.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %341 : Tensor = aten::add(%out.43, %335, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %342 : Tensor = aten::relu(%341), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1/__module.kp_detector.fg_encoder.layer3.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 (CaskConvolution)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - CaskConvolution has no valid tactics for this config, skipping
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %input.309 : Tensor = aten::_convolution(%338, %self.kp_detector.fg_encoder.layer3.1.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1/__module.kp_detector.fg_encoder.layer3.1.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.43 : Tensor = aten::batch_norm(%input.309, %self.kp_detector.fg_encoder.layer3.1.bn2.weight, %self.kp_detector.fg_encoder.layer3.1.bn2.bias, %self.kp_detector.fg_encoder.layer3.1.bn2.running_mean, %self.kp_detector.fg_encoder.layer3.1.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1/__module.kp_detector.fg_encoder.layer3.1.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %341 : Tensor = aten::add(%out.43, %335, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %342 : Tensor = aten::relu(%341), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1/__module.kp_detector.fg_encoder.layer3.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 (CaskFlattenConvolution)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - CaskFlattenConvolution has no valid tactics for this config, skipping
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Half(16384,1:4,1024,64), Half(16384,1:4,1024,64) -> Half(16384,1:4,1024,64) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %input.309 : Tensor = aten::_convolution(%338, %self.kp_detector.fg_encoder.layer3.1.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1/__module.kp_detector.fg_encoder.layer3.1.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.43 : Tensor = aten::batch_norm(%input.309, %self.kp_detector.fg_encoder.layer3.1.bn2.weight, %self.kp_detector.fg_encoder.layer3.1.bn2.bias, %self.kp_detector.fg_encoder.layer3.1.bn2.running_mean, %self.kp_detector.fg_encoder.layer3.1.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1/__module.kp_detector.fg_encoder.layer3.1.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %341 : Tensor = aten::add(%out.43, %335, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %342 : Tensor = aten::relu(%341), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1/__module.kp_detector.fg_encoder.layer3.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 (CaskConvolution)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - CaskConvolution has no valid tactics for this config, skipping
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %input.309 : Tensor = aten::_convolution(%338, %self.kp_detector.fg_encoder.layer3.1.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1/__module.kp_detector.fg_encoder.layer3.1.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.43 : Tensor = aten::batch_norm(%input.309, %self.kp_detector.fg_encoder.layer3.1.bn2.weight, %self.kp_detector.fg_encoder.layer3.1.bn2.bias, %self.kp_detector.fg_encoder.layer3.1.bn2.running_mean, %self.kp_detector.fg_encoder.layer3.1.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1/__module.kp_detector.fg_encoder.layer3.1.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %341 : Tensor = aten::add(%out.43, %335, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %342 : Tensor = aten::relu(%341), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1/__module.kp_detector.fg_encoder.layer3.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 (CaskFlattenConvolution)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - CaskFlattenConvolution has no valid tactics for this config, skipping
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Half(8192,1:8,512,32), Float(65536,256,16,1) -> Float(65536,256,16,1) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %input.309 : Tensor = aten::_convolution(%338, %self.kp_detector.fg_encoder.layer3.1.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1/__module.kp_detector.fg_encoder.layer3.1.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.43 : Tensor = aten::batch_norm(%input.309, %self.kp_detector.fg_encoder.layer3.1.bn2.weight, %self.kp_detector.fg_encoder.layer3.1.bn2.bias, %self.kp_detector.fg_encoder.layer3.1.bn2.running_mean, %self.kp_detector.fg_encoder.layer3.1.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1/__module.kp_detector.fg_encoder.layer3.1.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %341 : Tensor = aten::add(%out.43, %335, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %342 : Tensor = aten::relu(%341), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1/__module.kp_detector.fg_encoder.layer3.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 (CaskConvolution)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - CaskConvolution has no valid tactics for this config, skipping
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %input.309 : Tensor = aten::_convolution(%338, %self.kp_detector.fg_encoder.layer3.1.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1/__module.kp_detector.fg_encoder.layer3.1.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.43 : Tensor = aten::batch_norm(%input.309, %self.kp_detector.fg_encoder.layer3.1.bn2.weight, %self.kp_detector.fg_encoder.layer3.1.bn2.bias, %self.kp_detector.fg_encoder.layer3.1.bn2.running_mean, %self.kp_detector.fg_encoder.layer3.1.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1/__module.kp_detector.fg_encoder.layer3.1.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %341 : Tensor = aten::add(%out.43, %335, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %342 : Tensor = aten::relu(%341), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1/__module.kp_detector.fg_encoder.layer3.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 (CaskFlattenConvolution)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - CaskFlattenConvolution has no valid tactics for this config, skipping
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Half(8192,1:8,512,32), Half(8192,1:8,512,32) -> Half(8192,1:8,512,32) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - =============== Computing costs for 
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Float(65536,256,16,1), Float(65536,256,16,1) -> Float(65536,256,16,1) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Float(65536,1,4096,256), Float(65536,1,4096,256) -> Float(65536,1,4096,256) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Float(16384,1:4,1024,64), Float(16384,1:4,1024,64) -> Float(16384,1:4,1024,64) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Half(65536,256,16,1), Half(65536,256,16,1) -> Half(65536,256,16,1) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Half(32768,256:2,16,1), Half(32768,256:2,16,1) -> Half(32768,256:2,16,1) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %input.195 : Tensor = aten::_convolution(%259, %self.kp_detector.fg_encoder.layer3.1.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1/__module.kp_detector.fg_encoder.layer3.1.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.27 : Tensor = aten::batch_norm(%input.195, %self.kp_detector.fg_encoder.layer3.1.bn2.weight, %self.kp_detector.fg_encoder.layer3.1.bn2.bias, %self.kp_detector.fg_encoder.layer3.1.bn2.running_mean, %self.kp_detector.fg_encoder.layer3.1.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1/__module.kp_detector.fg_encoder.layer3.1.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %262 : Tensor = aten::add(%out.27, %256, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %263 : Tensor = aten::relu(%262), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1/__module.kp_detector.fg_encoder.layer3.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 (CaskConvolution)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - CaskConvolution has no valid tactics for this config, skipping
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %input.195 : Tensor = aten::_convolution(%259, %self.kp_detector.fg_encoder.layer3.1.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1/__module.kp_detector.fg_encoder.layer3.1.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.27 : Tensor = aten::batch_norm(%input.195, %self.kp_detector.fg_encoder.layer3.1.bn2.weight, %self.kp_detector.fg_encoder.layer3.1.bn2.bias, %self.kp_detector.fg_encoder.layer3.1.bn2.running_mean, %self.kp_detector.fg_encoder.layer3.1.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1/__module.kp_detector.fg_encoder.layer3.1.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %262 : Tensor = aten::add(%out.27, %256, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %263 : Tensor = aten::relu(%262), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1/__module.kp_detector.fg_encoder.layer3.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 (CaskFlattenConvolution)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - CaskFlattenConvolution has no valid tactics for this config, skipping
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Half(16384,1:4,1024,64), Half(16384,1:4,1024,64) -> Half(16384,1:4,1024,64) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %input.195 : Tensor = aten::_convolution(%259, %self.kp_detector.fg_encoder.layer3.1.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1/__module.kp_detector.fg_encoder.layer3.1.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.27 : Tensor = aten::batch_norm(%input.195, %self.kp_detector.fg_encoder.layer3.1.bn2.weight, %self.kp_detector.fg_encoder.layer3.1.bn2.bias, %self.kp_detector.fg_encoder.layer3.1.bn2.running_mean, %self.kp_detector.fg_encoder.layer3.1.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1/__module.kp_detector.fg_encoder.layer3.1.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %262 : Tensor = aten::add(%out.27, %256, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %263 : Tensor = aten::relu(%262), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1/__module.kp_detector.fg_encoder.layer3.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 (CaskConvolution)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - CaskConvolution has no valid tactics for this config, skipping
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %input.195 : Tensor = aten::_convolution(%259, %self.kp_detector.fg_encoder.layer3.1.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1/__module.kp_detector.fg_encoder.layer3.1.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.27 : Tensor = aten::batch_norm(%input.195, %self.kp_detector.fg_encoder.layer3.1.bn2.weight, %self.kp_detector.fg_encoder.layer3.1.bn2.bias, %self.kp_detector.fg_encoder.layer3.1.bn2.running_mean, %self.kp_detector.fg_encoder.layer3.1.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1/__module.kp_detector.fg_encoder.layer3.1.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %262 : Tensor = aten::add(%out.27, %256, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %263 : Tensor = aten::relu(%262), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1/__module.kp_detector.fg_encoder.layer3.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 (CaskFlattenConvolution)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - CaskFlattenConvolution has no valid tactics for this config, skipping
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Half(8192,1:8,512,32), Float(65536,256,16,1) -> Float(65536,256,16,1) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %input.195 : Tensor = aten::_convolution(%259, %self.kp_detector.fg_encoder.layer3.1.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1/__module.kp_detector.fg_encoder.layer3.1.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.27 : Tensor = aten::batch_norm(%input.195, %self.kp_detector.fg_encoder.layer3.1.bn2.weight, %self.kp_detector.fg_encoder.layer3.1.bn2.bias, %self.kp_detector.fg_encoder.layer3.1.bn2.running_mean, %self.kp_detector.fg_encoder.layer3.1.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1/__module.kp_detector.fg_encoder.layer3.1.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %262 : Tensor = aten::add(%out.27, %256, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %263 : Tensor = aten::relu(%262), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1/__module.kp_detector.fg_encoder.layer3.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 (CaskConvolution)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - CaskConvolution has no valid tactics for this config, skipping
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %input.195 : Tensor = aten::_convolution(%259, %self.kp_detector.fg_encoder.layer3.1.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1/__module.kp_detector.fg_encoder.layer3.1.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.27 : Tensor = aten::batch_norm(%input.195, %self.kp_detector.fg_encoder.layer3.1.bn2.weight, %self.kp_detector.fg_encoder.layer3.1.bn2.bias, %self.kp_detector.fg_encoder.layer3.1.bn2.running_mean, %self.kp_detector.fg_encoder.layer3.1.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1/__module.kp_detector.fg_encoder.layer3.1.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %262 : Tensor = aten::add(%out.27, %256, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %263 : Tensor = aten::relu(%262), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1/__module.kp_detector.fg_encoder.layer3.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 (CaskFlattenConvolution)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - CaskFlattenConvolution has no valid tactics for this config, skipping
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Half(8192,1:8,512,32), Half(8192,1:8,512,32) -> Half(8192,1:8,512,32) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - =============== Computing costs for 
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Float(65536,256,16,1) -> Float(32768,64,8,1) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Float(65536,1,4096,256) -> Float(32768,1,4096,512) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Float(16384,1:4,1024,64) -> Float(8192,1:4,1024,128) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Half(65536,256,16,1) -> Half(32768,64,8,1) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Half(32768,256:2,16,1) -> Half(16384,64:2,8,1) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %input.201 : Tensor = aten::_convolution(%263, %self.kp_detector.fg_encoder.layer4.0.conv1.weight, %5, %25, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.203 : Tensor = aten::batch_norm(%input.201, %self.kp_detector.fg_encoder.layer4.0.bn1.weight, %self.kp_detector.fg_encoder.layer4.0.bn1.bias, %self.kp_detector.fg_encoder.layer4.0.bn1.running_mean, %self.kp_detector.fg_encoder.layer4.0.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %266 : Tensor = aten::relu(%input.203), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 (CaskConvolution)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - CaskConvolution has no valid tactics for this config, skipping
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %input.201 : Tensor = aten::_convolution(%263, %self.kp_detector.fg_encoder.layer4.0.conv1.weight, %5, %25, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.203 : Tensor = aten::batch_norm(%input.201, %self.kp_detector.fg_encoder.layer4.0.bn1.weight, %self.kp_detector.fg_encoder.layer4.0.bn1.bias, %self.kp_detector.fg_encoder.layer4.0.bn1.running_mean, %self.kp_detector.fg_encoder.layer4.0.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %266 : Tensor = aten::relu(%input.203), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 (CaskFlattenConvolution)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - CaskFlattenConvolution has no valid tactics for this config, skipping
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Half(16384,1:4,1024,64) -> Half(8192,1:4,1024,128) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %input.201 : Tensor = aten::_convolution(%263, %self.kp_detector.fg_encoder.layer4.0.conv1.weight, %5, %25, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.203 : Tensor = aten::batch_norm(%input.201, %self.kp_detector.fg_encoder.layer4.0.bn1.weight, %self.kp_detector.fg_encoder.layer4.0.bn1.bias, %self.kp_detector.fg_encoder.layer4.0.bn1.running_mean, %self.kp_detector.fg_encoder.layer4.0.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %266 : Tensor = aten::relu(%input.203), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 (CaskConvolution)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - CaskConvolution has no valid tactics for this config, skipping
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %input.201 : Tensor = aten::_convolution(%263, %self.kp_detector.fg_encoder.layer4.0.conv1.weight, %5, %25, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.203 : Tensor = aten::batch_norm(%input.201, %self.kp_detector.fg_encoder.layer4.0.bn1.weight, %self.kp_detector.fg_encoder.layer4.0.bn1.bias, %self.kp_detector.fg_encoder.layer4.0.bn1.running_mean, %self.kp_detector.fg_encoder.layer4.0.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %266 : Tensor = aten::relu(%input.203), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 (CaskFlattenConvolution)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - CaskFlattenConvolution has no valid tactics for this config, skipping
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Half(8192,1:8,512,32) -> Float(32768,64,8,1) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %input.201 : Tensor = aten::_convolution(%263, %self.kp_detector.fg_encoder.layer4.0.conv1.weight, %5, %25, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.203 : Tensor = aten::batch_norm(%input.201, %self.kp_detector.fg_encoder.layer4.0.bn1.weight, %self.kp_detector.fg_encoder.layer4.0.bn1.bias, %self.kp_detector.fg_encoder.layer4.0.bn1.running_mean, %self.kp_detector.fg_encoder.layer4.0.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %266 : Tensor = aten::relu(%input.203), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 (CaskConvolution)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - CaskConvolution has no valid tactics for this config, skipping
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %input.201 : Tensor = aten::_convolution(%263, %self.kp_detector.fg_encoder.layer4.0.conv1.weight, %5, %25, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.203 : Tensor = aten::batch_norm(%input.201, %self.kp_detector.fg_encoder.layer4.0.bn1.weight, %self.kp_detector.fg_encoder.layer4.0.bn1.bias, %self.kp_detector.fg_encoder.layer4.0.bn1.running_mean, %self.kp_detector.fg_encoder.layer4.0.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %266 : Tensor = aten::relu(%input.203), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 (CaskFlattenConvolution)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - CaskFlattenConvolution has no valid tactics for this config, skipping
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Half(8192,1:8,512,32) -> Half(4096,1:8,512,64) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - =============== Computing costs for 
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Float(65536,256,16,1) -> Float(32768,64,8,1) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Float(65536,1,4096,256) -> Float(32768,1,4096,512) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Float(16384,1:4,1024,64) -> Float(8192,1:4,1024,128) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Half(65536,256,16,1) -> Half(32768,64,8,1) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Half(32768,256:2,16,1) -> Half(16384,64:2,8,1) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %input.315 : Tensor = aten::_convolution(%342, %self.kp_detector.fg_encoder.layer4.0.conv1.weight, %5, %25, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.317 : Tensor = aten::batch_norm(%input.315, %self.kp_detector.fg_encoder.layer4.0.bn1.weight, %self.kp_detector.fg_encoder.layer4.0.bn1.bias, %self.kp_detector.fg_encoder.layer4.0.bn1.running_mean, %self.kp_detector.fg_encoder.layer4.0.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %345 : Tensor = aten::relu(%input.317), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 (CaskConvolution)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - CaskConvolution has no valid tactics for this config, skipping
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %input.315 : Tensor = aten::_convolution(%342, %self.kp_detector.fg_encoder.layer4.0.conv1.weight, %5, %25, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.317 : Tensor = aten::batch_norm(%input.315, %self.kp_detector.fg_encoder.layer4.0.bn1.weight, %self.kp_detector.fg_encoder.layer4.0.bn1.bias, %self.kp_detector.fg_encoder.layer4.0.bn1.running_mean, %self.kp_detector.fg_encoder.layer4.0.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %345 : Tensor = aten::relu(%input.317), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 (CaskFlattenConvolution)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - CaskFlattenConvolution has no valid tactics for this config, skipping
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Half(16384,1:4,1024,64) -> Half(8192,1:4,1024,128) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %input.315 : Tensor = aten::_convolution(%342, %self.kp_detector.fg_encoder.layer4.0.conv1.weight, %5, %25, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.317 : Tensor = aten::batch_norm(%input.315, %self.kp_detector.fg_encoder.layer4.0.bn1.weight, %self.kp_detector.fg_encoder.layer4.0.bn1.bias, %self.kp_detector.fg_encoder.layer4.0.bn1.running_mean, %self.kp_detector.fg_encoder.layer4.0.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %345 : Tensor = aten::relu(%input.317), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 (CaskConvolution)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - CaskConvolution has no valid tactics for this config, skipping
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %input.315 : Tensor = aten::_convolution(%342, %self.kp_detector.fg_encoder.layer4.0.conv1.weight, %5, %25, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.317 : Tensor = aten::batch_norm(%input.315, %self.kp_detector.fg_encoder.layer4.0.bn1.weight, %self.kp_detector.fg_encoder.layer4.0.bn1.bias, %self.kp_detector.fg_encoder.layer4.0.bn1.running_mean, %self.kp_detector.fg_encoder.layer4.0.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %345 : Tensor = aten::relu(%input.317), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 (CaskFlattenConvolution)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - CaskFlattenConvolution has no valid tactics for this config, skipping
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Half(8192,1:8,512,32) -> Float(32768,64,8,1) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %input.315 : Tensor = aten::_convolution(%342, %self.kp_detector.fg_encoder.layer4.0.conv1.weight, %5, %25, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.317 : Tensor = aten::batch_norm(%input.315, %self.kp_detector.fg_encoder.layer4.0.bn1.weight, %self.kp_detector.fg_encoder.layer4.0.bn1.bias, %self.kp_detector.fg_encoder.layer4.0.bn1.running_mean, %self.kp_detector.fg_encoder.layer4.0.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %345 : Tensor = aten::relu(%input.317), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 (CaskConvolution)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - CaskConvolution has no valid tactics for this config, skipping
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %input.315 : Tensor = aten::_convolution(%342, %self.kp_detector.fg_encoder.layer4.0.conv1.weight, %5, %25, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.317 : Tensor = aten::batch_norm(%input.315, %self.kp_detector.fg_encoder.layer4.0.bn1.weight, %self.kp_detector.fg_encoder.layer4.0.bn1.bias, %self.kp_detector.fg_encoder.layer4.0.bn1.running_mean, %self.kp_detector.fg_encoder.layer4.0.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %345 : Tensor = aten::relu(%input.317), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 (CaskFlattenConvolution)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - CaskFlattenConvolution has no valid tactics for this config, skipping
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Half(8192,1:8,512,32) -> Half(4096,1:8,512,64) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - =============== Computing costs for 
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Float(32768,64,8,1) -> Float(32768,64,8,1) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Float(32768,1,4096,512) -> Float(32768,1,4096,512) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Float(8192,1:4,1024,128) -> Float(8192,1:4,1024,128) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Half(32768,64,8,1) -> Half(32768,64,8,1) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Half(16384,64:2,8,1) -> Half(16384,64:2,8,1) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %input.321 : Tensor = aten::_convolution(%345, %self.kp_detector.fg_encoder.layer4.0.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.45 : Tensor = aten::batch_norm(%input.321, %self.kp_detector.fg_encoder.layer4.0.bn2.weight, %self.kp_detector.fg_encoder.layer4.0.bn2.bias, %self.kp_detector.fg_encoder.layer4.0.bn2.running_mean, %self.kp_detector.fg_encoder.layer4.0.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 (CaskConvolution)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - CaskConvolution has no valid tactics for this config, skipping
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %input.321 : Tensor = aten::_convolution(%345, %self.kp_detector.fg_encoder.layer4.0.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.45 : Tensor = aten::batch_norm(%input.321, %self.kp_detector.fg_encoder.layer4.0.bn2.weight, %self.kp_detector.fg_encoder.layer4.0.bn2.bias, %self.kp_detector.fg_encoder.layer4.0.bn2.running_mean, %self.kp_detector.fg_encoder.layer4.0.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 (CaskFlattenConvolution)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - CaskFlattenConvolution has no valid tactics for this config, skipping
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Half(8192,1:4,1024,128) -> Half(8192,1:4,1024,128) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %input.321 : Tensor = aten::_convolution(%345, %self.kp_detector.fg_encoder.layer4.0.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.45 : Tensor = aten::batch_norm(%input.321, %self.kp_detector.fg_encoder.layer4.0.bn2.weight, %self.kp_detector.fg_encoder.layer4.0.bn2.bias, %self.kp_detector.fg_encoder.layer4.0.bn2.running_mean, %self.kp_detector.fg_encoder.layer4.0.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 (CaskConvolution)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - CaskConvolution has no valid tactics for this config, skipping
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %input.321 : Tensor = aten::_convolution(%345, %self.kp_detector.fg_encoder.layer4.0.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.45 : Tensor = aten::batch_norm(%input.321, %self.kp_detector.fg_encoder.layer4.0.bn2.weight, %self.kp_detector.fg_encoder.layer4.0.bn2.bias, %self.kp_detector.fg_encoder.layer4.0.bn2.running_mean, %self.kp_detector.fg_encoder.layer4.0.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 (CaskFlattenConvolution)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - CaskFlattenConvolution has no valid tactics for this config, skipping
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Half(4096,1:8,512,64) -> Float(32768,64,8,1) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %input.321 : Tensor = aten::_convolution(%345, %self.kp_detector.fg_encoder.layer4.0.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.45 : Tensor = aten::batch_norm(%input.321, %self.kp_detector.fg_encoder.layer4.0.bn2.weight, %self.kp_detector.fg_encoder.layer4.0.bn2.bias, %self.kp_detector.fg_encoder.layer4.0.bn2.running_mean, %self.kp_detector.fg_encoder.layer4.0.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 (CaskConvolution)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - CaskConvolution has no valid tactics for this config, skipping
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %input.321 : Tensor = aten::_convolution(%345, %self.kp_detector.fg_encoder.layer4.0.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.45 : Tensor = aten::batch_norm(%input.321, %self.kp_detector.fg_encoder.layer4.0.bn2.weight, %self.kp_detector.fg_encoder.layer4.0.bn2.bias, %self.kp_detector.fg_encoder.layer4.0.bn2.running_mean, %self.kp_detector.fg_encoder.layer4.0.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 (CaskFlattenConvolution)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - CaskFlattenConvolution has no valid tactics for this config, skipping
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Half(4096,1:8,512,64) -> Half(4096,1:8,512,64) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - =============== Computing costs for 
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Float(32768,64,8,1) -> Float(32768,64,8,1) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Float(32768,1,4096,512) -> Float(32768,1,4096,512) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Float(8192,1:4,1024,128) -> Float(8192,1:4,1024,128) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Half(32768,64,8,1) -> Half(32768,64,8,1) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Half(16384,64:2,8,1) -> Half(16384,64:2,8,1) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %input.207 : Tensor = aten::_convolution(%266, %self.kp_detector.fg_encoder.layer4.0.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.29 : Tensor = aten::batch_norm(%input.207, %self.kp_detector.fg_encoder.layer4.0.bn2.weight, %self.kp_detector.fg_encoder.layer4.0.bn2.bias, %self.kp_detector.fg_encoder.layer4.0.bn2.running_mean, %self.kp_detector.fg_encoder.layer4.0.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 (CaskConvolution)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - CaskConvolution has no valid tactics for this config, skipping
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %input.207 : Tensor = aten::_convolution(%266, %self.kp_detector.fg_encoder.layer4.0.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.29 : Tensor = aten::batch_norm(%input.207, %self.kp_detector.fg_encoder.layer4.0.bn2.weight, %self.kp_detector.fg_encoder.layer4.0.bn2.bias, %self.kp_detector.fg_encoder.layer4.0.bn2.running_mean, %self.kp_detector.fg_encoder.layer4.0.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 (CaskFlattenConvolution)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - CaskFlattenConvolution has no valid tactics for this config, skipping
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Half(8192,1:4,1024,128) -> Half(8192,1:4,1024,128) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %input.207 : Tensor = aten::_convolution(%266, %self.kp_detector.fg_encoder.layer4.0.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.29 : Tensor = aten::batch_norm(%input.207, %self.kp_detector.fg_encoder.layer4.0.bn2.weight, %self.kp_detector.fg_encoder.layer4.0.bn2.bias, %self.kp_detector.fg_encoder.layer4.0.bn2.running_mean, %self.kp_detector.fg_encoder.layer4.0.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 (CaskConvolution)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - CaskConvolution has no valid tactics for this config, skipping
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %input.207 : Tensor = aten::_convolution(%266, %self.kp_detector.fg_encoder.layer4.0.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.29 : Tensor = aten::batch_norm(%input.207, %self.kp_detector.fg_encoder.layer4.0.bn2.weight, %self.kp_detector.fg_encoder.layer4.0.bn2.bias, %self.kp_detector.fg_encoder.layer4.0.bn2.running_mean, %self.kp_detector.fg_encoder.layer4.0.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 (CaskFlattenConvolution)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - CaskFlattenConvolution has no valid tactics for this config, skipping
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Half(4096,1:8,512,64) -> Float(32768,64,8,1) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %input.207 : Tensor = aten::_convolution(%266, %self.kp_detector.fg_encoder.layer4.0.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.29 : Tensor = aten::batch_norm(%input.207, %self.kp_detector.fg_encoder.layer4.0.bn2.weight, %self.kp_detector.fg_encoder.layer4.0.bn2.bias, %self.kp_detector.fg_encoder.layer4.0.bn2.running_mean, %self.kp_detector.fg_encoder.layer4.0.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 (CaskConvolution)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - CaskConvolution has no valid tactics for this config, skipping
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %input.207 : Tensor = aten::_convolution(%266, %self.kp_detector.fg_encoder.layer4.0.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.29 : Tensor = aten::batch_norm(%input.207, %self.kp_detector.fg_encoder.layer4.0.bn2.weight, %self.kp_detector.fg_encoder.layer4.0.bn2.bias, %self.kp_detector.fg_encoder.layer4.0.bn2.running_mean, %self.kp_detector.fg_encoder.layer4.0.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 (CaskFlattenConvolution)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - CaskFlattenConvolution has no valid tactics for this config, skipping
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Half(4096,1:8,512,64) -> Half(4096,1:8,512,64) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - =============== Computing costs for 
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Float(65536,256,16,1), Float(32768,64,8,1) -> Float(32768,64,8,1) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Float(65536,1,4096,256), Float(32768,1,4096,512) -> Float(32768,1,4096,512) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Float(16384,1:4,1024,64), Float(8192,1:4,1024,128) -> Float(8192,1:4,1024,128) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Half(65536,256,16,1), Half(32768,64,8,1) -> Half(32768,64,8,1) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Half(32768,256:2,16,1), Half(16384,64:2,8,1) -> Half(16384,64:2,8,1) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %input.323 : Tensor = aten::_convolution(%342, %self.kp_detector.fg_encoder.layer4.0.downsample.0.weight, %5, %25, %28, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.downsample/__module.kp_detector.fg_encoder.layer4.0.downsample.0 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %identity : Tensor = aten::batch_norm(%input.323, %self.kp_detector.fg_encoder.layer4.0.downsample.1.weight, %self.kp_detector.fg_encoder.layer4.0.bn2.bias, %self.kp_detector.fg_encoder.layer4.0.downsample.1.running_mean, %self.kp_detector.fg_encoder.layer4.0.downsample.1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.downsample/__module.kp_detector.fg_encoder.layer4.0.downsample.1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %350 : Tensor = aten::add(%out.45, %identity, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %351 : Tensor = aten::relu(%350), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 (CaskConvolution)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - CaskConvolution has no valid tactics for this config, skipping
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %input.323 : Tensor = aten::_convolution(%342, %self.kp_detector.fg_encoder.layer4.0.downsample.0.weight, %5, %25, %28, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.downsample/__module.kp_detector.fg_encoder.layer4.0.downsample.0 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %identity : Tensor = aten::batch_norm(%input.323, %self.kp_detector.fg_encoder.layer4.0.downsample.1.weight, %self.kp_detector.fg_encoder.layer4.0.bn2.bias, %self.kp_detector.fg_encoder.layer4.0.downsample.1.running_mean, %self.kp_detector.fg_encoder.layer4.0.downsample.1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.downsample/__module.kp_detector.fg_encoder.layer4.0.downsample.1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %350 : Tensor = aten::add(%out.45, %identity, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %351 : Tensor = aten::relu(%350), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 (CaskFlattenConvolution)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - CaskFlattenConvolution has no valid tactics for this config, skipping
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Half(16384,1:4,1024,64), Half(8192,1:4,1024,128) -> Half(8192,1:4,1024,128) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %input.323 : Tensor = aten::_convolution(%342, %self.kp_detector.fg_encoder.layer4.0.downsample.0.weight, %5, %25, %28, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.downsample/__module.kp_detector.fg_encoder.layer4.0.downsample.0 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %identity : Tensor = aten::batch_norm(%input.323, %self.kp_detector.fg_encoder.layer4.0.downsample.1.weight, %self.kp_detector.fg_encoder.layer4.0.bn2.bias, %self.kp_detector.fg_encoder.layer4.0.downsample.1.running_mean, %self.kp_detector.fg_encoder.layer4.0.downsample.1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.downsample/__module.kp_detector.fg_encoder.layer4.0.downsample.1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %350 : Tensor = aten::add(%out.45, %identity, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %351 : Tensor = aten::relu(%350), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 (CaskConvolution)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - CaskConvolution has no valid tactics for this config, skipping
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %input.323 : Tensor = aten::_convolution(%342, %self.kp_detector.fg_encoder.layer4.0.downsample.0.weight, %5, %25, %28, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.downsample/__module.kp_detector.fg_encoder.layer4.0.downsample.0 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %identity : Tensor = aten::batch_norm(%input.323, %self.kp_detector.fg_encoder.layer4.0.downsample.1.weight, %self.kp_detector.fg_encoder.layer4.0.bn2.bias, %self.kp_detector.fg_encoder.layer4.0.downsample.1.running_mean, %self.kp_detector.fg_encoder.layer4.0.downsample.1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.downsample/__module.kp_detector.fg_encoder.layer4.0.downsample.1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %350 : Tensor = aten::add(%out.45, %identity, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %351 : Tensor = aten::relu(%350), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 (CaskFlattenConvolution)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - CaskFlattenConvolution has no valid tactics for this config, skipping
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Half(8192,1:8,512,32), Float(32768,64,8,1) -> Float(32768,64,8,1) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %input.323 : Tensor = aten::_convolution(%342, %self.kp_detector.fg_encoder.layer4.0.downsample.0.weight, %5, %25, %28, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.downsample/__module.kp_detector.fg_encoder.layer4.0.downsample.0 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %identity : Tensor = aten::batch_norm(%input.323, %self.kp_detector.fg_encoder.layer4.0.downsample.1.weight, %self.kp_detector.fg_encoder.layer4.0.bn2.bias, %self.kp_detector.fg_encoder.layer4.0.downsample.1.running_mean, %self.kp_detector.fg_encoder.layer4.0.downsample.1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.downsample/__module.kp_detector.fg_encoder.layer4.0.downsample.1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %350 : Tensor = aten::add(%out.45, %identity, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %351 : Tensor = aten::relu(%350), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 (CaskConvolution)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - CaskConvolution has no valid tactics for this config, skipping
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %input.323 : Tensor = aten::_convolution(%342, %self.kp_detector.fg_encoder.layer4.0.downsample.0.weight, %5, %25, %28, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.downsample/__module.kp_detector.fg_encoder.layer4.0.downsample.0 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %identity : Tensor = aten::batch_norm(%input.323, %self.kp_detector.fg_encoder.layer4.0.downsample.1.weight, %self.kp_detector.fg_encoder.layer4.0.bn2.bias, %self.kp_detector.fg_encoder.layer4.0.downsample.1.running_mean, %self.kp_detector.fg_encoder.layer4.0.downsample.1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.downsample/__module.kp_detector.fg_encoder.layer4.0.downsample.1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %350 : Tensor = aten::add(%out.45, %identity, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %351 : Tensor = aten::relu(%350), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 (CaskFlattenConvolution)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - CaskFlattenConvolution has no valid tactics for this config, skipping
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Half(8192,1:8,512,32), Half(4096,1:8,512,64) -> Half(4096,1:8,512,64) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - =============== Computing costs for 
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Float(65536,256,16,1), Float(32768,64,8,1) -> Float(32768,64,8,1) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Float(65536,1,4096,256), Float(32768,1,4096,512) -> Float(32768,1,4096,512) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Float(16384,1:4,1024,64), Float(8192,1:4,1024,128) -> Float(8192,1:4,1024,128) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Half(65536,256,16,1), Half(32768,64,8,1) -> Half(32768,64,8,1) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Half(32768,256:2,16,1), Half(16384,64:2,8,1) -> Half(16384,64:2,8,1) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %input.209 : Tensor = aten::_convolution(%263, %self.kp_detector.fg_encoder.layer4.0.downsample.0.weight, %5, %25, %28, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.downsample/__module.kp_detector.fg_encoder.layer4.0.downsample.0 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %identity.11 : Tensor = aten::batch_norm(%input.209, %self.kp_detector.fg_encoder.layer4.0.downsample.1.weight, %self.kp_detector.fg_encoder.layer4.0.bn2.bias, %self.kp_detector.fg_encoder.layer4.0.downsample.1.running_mean, %self.kp_detector.fg_encoder.layer4.0.downsample.1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.downsample/__module.kp_detector.fg_encoder.layer4.0.downsample.1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %271 : Tensor = aten::add(%out.29, %identity.11, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %272 : Tensor = aten::relu(%271), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 (CaskConvolution)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - CaskConvolution has no valid tactics for this config, skipping
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %input.209 : Tensor = aten::_convolution(%263, %self.kp_detector.fg_encoder.layer4.0.downsample.0.weight, %5, %25, %28, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.downsample/__module.kp_detector.fg_encoder.layer4.0.downsample.0 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %identity.11 : Tensor = aten::batch_norm(%input.209, %self.kp_detector.fg_encoder.layer4.0.downsample.1.weight, %self.kp_detector.fg_encoder.layer4.0.bn2.bias, %self.kp_detector.fg_encoder.layer4.0.downsample.1.running_mean, %self.kp_detector.fg_encoder.layer4.0.downsample.1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.downsample/__module.kp_detector.fg_encoder.layer4.0.downsample.1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %271 : Tensor = aten::add(%out.29, %identity.11, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %272 : Tensor = aten::relu(%271), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 (CaskFlattenConvolution)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - CaskFlattenConvolution has no valid tactics for this config, skipping
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Half(16384,1:4,1024,64), Half(8192,1:4,1024,128) -> Half(8192,1:4,1024,128) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %input.209 : Tensor = aten::_convolution(%263, %self.kp_detector.fg_encoder.layer4.0.downsample.0.weight, %5, %25, %28, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.downsample/__module.kp_detector.fg_encoder.layer4.0.downsample.0 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %identity.11 : Tensor = aten::batch_norm(%input.209, %self.kp_detector.fg_encoder.layer4.0.downsample.1.weight, %self.kp_detector.fg_encoder.layer4.0.bn2.bias, %self.kp_detector.fg_encoder.layer4.0.downsample.1.running_mean, %self.kp_detector.fg_encoder.layer4.0.downsample.1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.downsample/__module.kp_detector.fg_encoder.layer4.0.downsample.1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %271 : Tensor = aten::add(%out.29, %identity.11, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %272 : Tensor = aten::relu(%271), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 (CaskConvolution)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - CaskConvolution has no valid tactics for this config, skipping
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %input.209 : Tensor = aten::_convolution(%263, %self.kp_detector.fg_encoder.layer4.0.downsample.0.weight, %5, %25, %28, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.downsample/__module.kp_detector.fg_encoder.layer4.0.downsample.0 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %identity.11 : Tensor = aten::batch_norm(%input.209, %self.kp_detector.fg_encoder.layer4.0.downsample.1.weight, %self.kp_detector.fg_encoder.layer4.0.bn2.bias, %self.kp_detector.fg_encoder.layer4.0.downsample.1.running_mean, %self.kp_detector.fg_encoder.layer4.0.downsample.1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.downsample/__module.kp_detector.fg_encoder.layer4.0.downsample.1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %271 : Tensor = aten::add(%out.29, %identity.11, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %272 : Tensor = aten::relu(%271), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 (CaskFlattenConvolution)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - CaskFlattenConvolution has no valid tactics for this config, skipping
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Half(8192,1:8,512,32), Float(32768,64,8,1) -> Float(32768,64,8,1) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %input.209 : Tensor = aten::_convolution(%263, %self.kp_detector.fg_encoder.layer4.0.downsample.0.weight, %5, %25, %28, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.downsample/__module.kp_detector.fg_encoder.layer4.0.downsample.0 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %identity.11 : Tensor = aten::batch_norm(%input.209, %self.kp_detector.fg_encoder.layer4.0.downsample.1.weight, %self.kp_detector.fg_encoder.layer4.0.bn2.bias, %self.kp_detector.fg_encoder.layer4.0.downsample.1.running_mean, %self.kp_detector.fg_encoder.layer4.0.downsample.1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.downsample/__module.kp_detector.fg_encoder.layer4.0.downsample.1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %271 : Tensor = aten::add(%out.29, %identity.11, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %272 : Tensor = aten::relu(%271), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 (CaskConvolution)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - CaskConvolution has no valid tactics for this config, skipping
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %input.209 : Tensor = aten::_convolution(%263, %self.kp_detector.fg_encoder.layer4.0.downsample.0.weight, %5, %25, %28, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.downsample/__module.kp_detector.fg_encoder.layer4.0.downsample.0 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %identity.11 : Tensor = aten::batch_norm(%input.209, %self.kp_detector.fg_encoder.layer4.0.downsample.1.weight, %self.kp_detector.fg_encoder.layer4.0.bn2.bias, %self.kp_detector.fg_encoder.layer4.0.downsample.1.running_mean, %self.kp_detector.fg_encoder.layer4.0.downsample.1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.downsample/__module.kp_detector.fg_encoder.layer4.0.downsample.1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %271 : Tensor = aten::add(%out.29, %identity.11, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %272 : Tensor = aten::relu(%271), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 (CaskFlattenConvolution)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - CaskFlattenConvolution has no valid tactics for this config, skipping
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Half(8192,1:8,512,32), Half(4096,1:8,512,64) -> Half(4096,1:8,512,64) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - =============== Computing costs for 
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Float(32768,64,8,1) -> Float(32768,64,8,1) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Float(32768,1,4096,512) -> Float(32768,1,4096,512) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Float(8192,1:4,1024,128) -> Float(8192,1:4,1024,128) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Half(32768,64,8,1) -> Half(32768,64,8,1) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Half(16384,64:2,8,1) -> Half(16384,64:2,8,1) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %input.215 : Tensor = aten::_convolution(%272, %self.kp_detector.fg_encoder.layer4.1.conv1.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1/__module.kp_detector.fg_encoder.layer4.1.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.217 : Tensor = aten::batch_norm(%input.215, %self.kp_detector.fg_encoder.layer4.1.bn1.weight, %self.kp_detector.fg_encoder.layer4.1.bn1.bias, %self.kp_detector.fg_encoder.layer4.1.bn1.running_mean, %self.kp_detector.fg_encoder.layer4.1.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1/__module.kp_detector.fg_encoder.layer4.1.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %275 : Tensor = aten::relu(%input.217), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1/__module.kp_detector.fg_encoder.layer4.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 (CaskConvolution)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - CaskConvolution has no valid tactics for this config, skipping
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %input.215 : Tensor = aten::_convolution(%272, %self.kp_detector.fg_encoder.layer4.1.conv1.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1/__module.kp_detector.fg_encoder.layer4.1.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.217 : Tensor = aten::batch_norm(%input.215, %self.kp_detector.fg_encoder.layer4.1.bn1.weight, %self.kp_detector.fg_encoder.layer4.1.bn1.bias, %self.kp_detector.fg_encoder.layer4.1.bn1.running_mean, %self.kp_detector.fg_encoder.layer4.1.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1/__module.kp_detector.fg_encoder.layer4.1.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %275 : Tensor = aten::relu(%input.217), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1/__module.kp_detector.fg_encoder.layer4.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 (CaskFlattenConvolution)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - CaskFlattenConvolution has no valid tactics for this config, skipping
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Half(8192,1:4,1024,128) -> Half(8192,1:4,1024,128) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %input.215 : Tensor = aten::_convolution(%272, %self.kp_detector.fg_encoder.layer4.1.conv1.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1/__module.kp_detector.fg_encoder.layer4.1.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.217 : Tensor = aten::batch_norm(%input.215, %self.kp_detector.fg_encoder.layer4.1.bn1.weight, %self.kp_detector.fg_encoder.layer4.1.bn1.bias, %self.kp_detector.fg_encoder.layer4.1.bn1.running_mean, %self.kp_detector.fg_encoder.layer4.1.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1/__module.kp_detector.fg_encoder.layer4.1.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %275 : Tensor = aten::relu(%input.217), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1/__module.kp_detector.fg_encoder.layer4.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 (CaskConvolution)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - CaskConvolution has no valid tactics for this config, skipping
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %input.215 : Tensor = aten::_convolution(%272, %self.kp_detector.fg_encoder.layer4.1.conv1.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1/__module.kp_detector.fg_encoder.layer4.1.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.217 : Tensor = aten::batch_norm(%input.215, %self.kp_detector.fg_encoder.layer4.1.bn1.weight, %self.kp_detector.fg_encoder.layer4.1.bn1.bias, %self.kp_detector.fg_encoder.layer4.1.bn1.running_mean, %self.kp_detector.fg_encoder.layer4.1.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1/__module.kp_detector.fg_encoder.layer4.1.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %275 : Tensor = aten::relu(%input.217), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1/__module.kp_detector.fg_encoder.layer4.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 (CaskFlattenConvolution)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - CaskFlattenConvolution has no valid tactics for this config, skipping
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Half(4096,1:8,512,64) -> Float(32768,64,8,1) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %input.215 : Tensor = aten::_convolution(%272, %self.kp_detector.fg_encoder.layer4.1.conv1.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1/__module.kp_detector.fg_encoder.layer4.1.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.217 : Tensor = aten::batch_norm(%input.215, %self.kp_detector.fg_encoder.layer4.1.bn1.weight, %self.kp_detector.fg_encoder.layer4.1.bn1.bias, %self.kp_detector.fg_encoder.layer4.1.bn1.running_mean, %self.kp_detector.fg_encoder.layer4.1.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1/__module.kp_detector.fg_encoder.layer4.1.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %275 : Tensor = aten::relu(%input.217), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1/__module.kp_detector.fg_encoder.layer4.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 (CaskConvolution)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - CaskConvolution has no valid tactics for this config, skipping
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %input.215 : Tensor = aten::_convolution(%272, %self.kp_detector.fg_encoder.layer4.1.conv1.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1/__module.kp_detector.fg_encoder.layer4.1.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.217 : Tensor = aten::batch_norm(%input.215, %self.kp_detector.fg_encoder.layer4.1.bn1.weight, %self.kp_detector.fg_encoder.layer4.1.bn1.bias, %self.kp_detector.fg_encoder.layer4.1.bn1.running_mean, %self.kp_detector.fg_encoder.layer4.1.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1/__module.kp_detector.fg_encoder.layer4.1.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %275 : Tensor = aten::relu(%input.217), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1/__module.kp_detector.fg_encoder.layer4.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 (CaskFlattenConvolution)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - CaskFlattenConvolution has no valid tactics for this config, skipping
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Half(4096,1:8,512,64) -> Half(4096,1:8,512,64) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - =============== Computing costs for 
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Float(32768,64,8,1) -> Float(32768,64,8,1) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Float(32768,1,4096,512) -> Float(32768,1,4096,512) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Float(8192,1:4,1024,128) -> Float(8192,1:4,1024,128) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Half(32768,64,8,1) -> Half(32768,64,8,1) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Half(16384,64:2,8,1) -> Half(16384,64:2,8,1) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %input.329 : Tensor = aten::_convolution(%351, %self.kp_detector.fg_encoder.layer4.1.conv1.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1/__module.kp_detector.fg_encoder.layer4.1.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.331 : Tensor = aten::batch_norm(%input.329, %self.kp_detector.fg_encoder.layer4.1.bn1.weight, %self.kp_detector.fg_encoder.layer4.1.bn1.bias, %self.kp_detector.fg_encoder.layer4.1.bn1.running_mean, %self.kp_detector.fg_encoder.layer4.1.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1/__module.kp_detector.fg_encoder.layer4.1.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %354 : Tensor = aten::relu(%input.331), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1/__module.kp_detector.fg_encoder.layer4.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 (CaskConvolution)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - CaskConvolution has no valid tactics for this config, skipping
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %input.329 : Tensor = aten::_convolution(%351, %self.kp_detector.fg_encoder.layer4.1.conv1.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1/__module.kp_detector.fg_encoder.layer4.1.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.331 : Tensor = aten::batch_norm(%input.329, %self.kp_detector.fg_encoder.layer4.1.bn1.weight, %self.kp_detector.fg_encoder.layer4.1.bn1.bias, %self.kp_detector.fg_encoder.layer4.1.bn1.running_mean, %self.kp_detector.fg_encoder.layer4.1.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1/__module.kp_detector.fg_encoder.layer4.1.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %354 : Tensor = aten::relu(%input.331), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1/__module.kp_detector.fg_encoder.layer4.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 (CaskFlattenConvolution)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - CaskFlattenConvolution has no valid tactics for this config, skipping
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Half(8192,1:4,1024,128) -> Half(8192,1:4,1024,128) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %input.329 : Tensor = aten::_convolution(%351, %self.kp_detector.fg_encoder.layer4.1.conv1.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1/__module.kp_detector.fg_encoder.layer4.1.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.331 : Tensor = aten::batch_norm(%input.329, %self.kp_detector.fg_encoder.layer4.1.bn1.weight, %self.kp_detector.fg_encoder.layer4.1.bn1.bias, %self.kp_detector.fg_encoder.layer4.1.bn1.running_mean, %self.kp_detector.fg_encoder.layer4.1.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1/__module.kp_detector.fg_encoder.layer4.1.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %354 : Tensor = aten::relu(%input.331), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1/__module.kp_detector.fg_encoder.layer4.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 (CaskConvolution)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - CaskConvolution has no valid tactics for this config, skipping
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %input.329 : Tensor = aten::_convolution(%351, %self.kp_detector.fg_encoder.layer4.1.conv1.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1/__module.kp_detector.fg_encoder.layer4.1.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.331 : Tensor = aten::batch_norm(%input.329, %self.kp_detector.fg_encoder.layer4.1.bn1.weight, %self.kp_detector.fg_encoder.layer4.1.bn1.bias, %self.kp_detector.fg_encoder.layer4.1.bn1.running_mean, %self.kp_detector.fg_encoder.layer4.1.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1/__module.kp_detector.fg_encoder.layer4.1.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %354 : Tensor = aten::relu(%input.331), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1/__module.kp_detector.fg_encoder.layer4.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 (CaskFlattenConvolution)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - CaskFlattenConvolution has no valid tactics for this config, skipping
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Half(4096,1:8,512,64) -> Float(32768,64,8,1) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %input.329 : Tensor = aten::_convolution(%351, %self.kp_detector.fg_encoder.layer4.1.conv1.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1/__module.kp_detector.fg_encoder.layer4.1.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.331 : Tensor = aten::batch_norm(%input.329, %self.kp_detector.fg_encoder.layer4.1.bn1.weight, %self.kp_detector.fg_encoder.layer4.1.bn1.bias, %self.kp_detector.fg_encoder.layer4.1.bn1.running_mean, %self.kp_detector.fg_encoder.layer4.1.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1/__module.kp_detector.fg_encoder.layer4.1.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %354 : Tensor = aten::relu(%input.331), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1/__module.kp_detector.fg_encoder.layer4.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 (CaskConvolution)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - CaskConvolution has no valid tactics for this config, skipping
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %input.329 : Tensor = aten::_convolution(%351, %self.kp_detector.fg_encoder.layer4.1.conv1.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1/__module.kp_detector.fg_encoder.layer4.1.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.331 : Tensor = aten::batch_norm(%input.329, %self.kp_detector.fg_encoder.layer4.1.bn1.weight, %self.kp_detector.fg_encoder.layer4.1.bn1.bias, %self.kp_detector.fg_encoder.layer4.1.bn1.running_mean, %self.kp_detector.fg_encoder.layer4.1.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1/__module.kp_detector.fg_encoder.layer4.1.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %354 : Tensor = aten::relu(%input.331), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1/__module.kp_detector.fg_encoder.layer4.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 (CaskFlattenConvolution)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - CaskFlattenConvolution has no valid tactics for this config, skipping
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Half(4096,1:8,512,64) -> Half(4096,1:8,512,64) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - =============== Computing costs for 
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Float(32768,64,8,1), Float(32768,64,8,1) -> Float(32768,64,8,1) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Float(32768,1,4096,512), Float(32768,1,4096,512) -> Float(32768,1,4096,512) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Float(8192,1:4,1024,128), Float(8192,1:4,1024,128) -> Float(8192,1:4,1024,128) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Half(32768,64,8,1), Half(32768,64,8,1) -> Half(32768,64,8,1) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Half(16384,64:2,8,1), Half(16384,64:2,8,1) -> Half(16384,64:2,8,1) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %input.335 : Tensor = aten::_convolution(%354, %self.kp_detector.fg_encoder.layer4.1.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1/__module.kp_detector.fg_encoder.layer4.1.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.47 : Tensor = aten::batch_norm(%input.335, %self.kp_detector.fg_encoder.layer4.1.bn2.weight, %self.kp_detector.fg_encoder.layer4.1.bn2.bias, %self.kp_detector.fg_encoder.layer4.1.bn2.running_mean, %self.kp_detector.fg_encoder.layer4.1.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1/__module.kp_detector.fg_encoder.layer4.1.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %357 : Tensor = aten::add(%out.47, %351, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %358 : Tensor = aten::relu(%357), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1/__module.kp_detector.fg_encoder.layer4.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 (CaskConvolution)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - CaskConvolution has no valid tactics for this config, skipping
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %input.335 : Tensor = aten::_convolution(%354, %self.kp_detector.fg_encoder.layer4.1.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1/__module.kp_detector.fg_encoder.layer4.1.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.47 : Tensor = aten::batch_norm(%input.335, %self.kp_detector.fg_encoder.layer4.1.bn2.weight, %self.kp_detector.fg_encoder.layer4.1.bn2.bias, %self.kp_detector.fg_encoder.layer4.1.bn2.running_mean, %self.kp_detector.fg_encoder.layer4.1.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1/__module.kp_detector.fg_encoder.layer4.1.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %357 : Tensor = aten::add(%out.47, %351, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %358 : Tensor = aten::relu(%357), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1/__module.kp_detector.fg_encoder.layer4.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 (CaskFlattenConvolution)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - CaskFlattenConvolution has no valid tactics for this config, skipping
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Half(8192,1:4,1024,128), Half(8192,1:4,1024,128) -> Half(8192,1:4,1024,128) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %input.335 : Tensor = aten::_convolution(%354, %self.kp_detector.fg_encoder.layer4.1.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1/__module.kp_detector.fg_encoder.layer4.1.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.47 : Tensor = aten::batch_norm(%input.335, %self.kp_detector.fg_encoder.layer4.1.bn2.weight, %self.kp_detector.fg_encoder.layer4.1.bn2.bias, %self.kp_detector.fg_encoder.layer4.1.bn2.running_mean, %self.kp_detector.fg_encoder.layer4.1.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1/__module.kp_detector.fg_encoder.layer4.1.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %357 : Tensor = aten::add(%out.47, %351, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %358 : Tensor = aten::relu(%357), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1/__module.kp_detector.fg_encoder.layer4.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 (CaskConvolution)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - CaskConvolution has no valid tactics for this config, skipping
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %input.335 : Tensor = aten::_convolution(%354, %self.kp_detector.fg_encoder.layer4.1.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1/__module.kp_detector.fg_encoder.layer4.1.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.47 : Tensor = aten::batch_norm(%input.335, %self.kp_detector.fg_encoder.layer4.1.bn2.weight, %self.kp_detector.fg_encoder.layer4.1.bn2.bias, %self.kp_detector.fg_encoder.layer4.1.bn2.running_mean, %self.kp_detector.fg_encoder.layer4.1.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1/__module.kp_detector.fg_encoder.layer4.1.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %357 : Tensor = aten::add(%out.47, %351, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %358 : Tensor = aten::relu(%357), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1/__module.kp_detector.fg_encoder.layer4.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 (CaskFlattenConvolution)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - CaskFlattenConvolution has no valid tactics for this config, skipping
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Half(4096,1:8,512,64), Float(32768,64,8,1) -> Float(32768,64,8,1) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %input.335 : Tensor = aten::_convolution(%354, %self.kp_detector.fg_encoder.layer4.1.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1/__module.kp_detector.fg_encoder.layer4.1.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.47 : Tensor = aten::batch_norm(%input.335, %self.kp_detector.fg_encoder.layer4.1.bn2.weight, %self.kp_detector.fg_encoder.layer4.1.bn2.bias, %self.kp_detector.fg_encoder.layer4.1.bn2.running_mean, %self.kp_detector.fg_encoder.layer4.1.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1/__module.kp_detector.fg_encoder.layer4.1.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %357 : Tensor = aten::add(%out.47, %351, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %358 : Tensor = aten::relu(%357), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1/__module.kp_detector.fg_encoder.layer4.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 (CaskConvolution)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - CaskConvolution has no valid tactics for this config, skipping
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %input.335 : Tensor = aten::_convolution(%354, %self.kp_detector.fg_encoder.layer4.1.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1/__module.kp_detector.fg_encoder.layer4.1.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.47 : Tensor = aten::batch_norm(%input.335, %self.kp_detector.fg_encoder.layer4.1.bn2.weight, %self.kp_detector.fg_encoder.layer4.1.bn2.bias, %self.kp_detector.fg_encoder.layer4.1.bn2.running_mean, %self.kp_detector.fg_encoder.layer4.1.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1/__module.kp_detector.fg_encoder.layer4.1.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %357 : Tensor = aten::add(%out.47, %351, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %358 : Tensor = aten::relu(%357), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1/__module.kp_detector.fg_encoder.layer4.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 (CaskFlattenConvolution)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - CaskFlattenConvolution has no valid tactics for this config, skipping
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Half(4096,1:8,512,64), Half(4096,1:8,512,64) -> Half(4096,1:8,512,64) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - =============== Computing costs for 
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Float(32768,64,8,1) -> Float(512,1,1,1) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Float(8192,1:4,1024,128) -> Float(128,1:4,128,128) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Half(32768,64,8,1) -> Half(512,1,1,1) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Half(16384,64:2,8,1) -> Half(256,1:2,1,1) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Half(4096,1:8,512,64) -> Half(64,1:8,64,64) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - =============== Computing costs for 
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Float(32768,64,8,1), Float(32768,64,8,1) -> Float(32768,64,8,1) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Float(32768,1,4096,512), Float(32768,1,4096,512) -> Float(32768,1,4096,512) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Float(8192,1:4,1024,128), Float(8192,1:4,1024,128) -> Float(8192,1:4,1024,128) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Half(32768,64,8,1), Half(32768,64,8,1) -> Half(32768,64,8,1) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Half(16384,64:2,8,1), Half(16384,64:2,8,1) -> Half(16384,64:2,8,1) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %input.221 : Tensor = aten::_convolution(%275, %self.kp_detector.fg_encoder.layer4.1.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1/__module.kp_detector.fg_encoder.layer4.1.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.31 : Tensor = aten::batch_norm(%input.221, %self.kp_detector.fg_encoder.layer4.1.bn2.weight, %self.kp_detector.fg_encoder.layer4.1.bn2.bias, %self.kp_detector.fg_encoder.layer4.1.bn2.running_mean, %self.kp_detector.fg_encoder.layer4.1.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1/__module.kp_detector.fg_encoder.layer4.1.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %278 : Tensor = aten::add(%out.31, %272, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %279 : Tensor = aten::relu(%278), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1/__module.kp_detector.fg_encoder.layer4.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 (CaskConvolution)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - CaskConvolution has no valid tactics for this config, skipping
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %input.221 : Tensor = aten::_convolution(%275, %self.kp_detector.fg_encoder.layer4.1.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1/__module.kp_detector.fg_encoder.layer4.1.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.31 : Tensor = aten::batch_norm(%input.221, %self.kp_detector.fg_encoder.layer4.1.bn2.weight, %self.kp_detector.fg_encoder.layer4.1.bn2.bias, %self.kp_detector.fg_encoder.layer4.1.bn2.running_mean, %self.kp_detector.fg_encoder.layer4.1.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1/__module.kp_detector.fg_encoder.layer4.1.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %278 : Tensor = aten::add(%out.31, %272, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %279 : Tensor = aten::relu(%278), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1/__module.kp_detector.fg_encoder.layer4.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 (CaskFlattenConvolution)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - CaskFlattenConvolution has no valid tactics for this config, skipping
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Half(8192,1:4,1024,128), Half(8192,1:4,1024,128) -> Half(8192,1:4,1024,128) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %input.221 : Tensor = aten::_convolution(%275, %self.kp_detector.fg_encoder.layer4.1.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1/__module.kp_detector.fg_encoder.layer4.1.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.31 : Tensor = aten::batch_norm(%input.221, %self.kp_detector.fg_encoder.layer4.1.bn2.weight, %self.kp_detector.fg_encoder.layer4.1.bn2.bias, %self.kp_detector.fg_encoder.layer4.1.bn2.running_mean, %self.kp_detector.fg_encoder.layer4.1.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1/__module.kp_detector.fg_encoder.layer4.1.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %278 : Tensor = aten::add(%out.31, %272, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %279 : Tensor = aten::relu(%278), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1/__module.kp_detector.fg_encoder.layer4.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 (CaskConvolution)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - CaskConvolution has no valid tactics for this config, skipping
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %input.221 : Tensor = aten::_convolution(%275, %self.kp_detector.fg_encoder.layer4.1.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1/__module.kp_detector.fg_encoder.layer4.1.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.31 : Tensor = aten::batch_norm(%input.221, %self.kp_detector.fg_encoder.layer4.1.bn2.weight, %self.kp_detector.fg_encoder.layer4.1.bn2.bias, %self.kp_detector.fg_encoder.layer4.1.bn2.running_mean, %self.kp_detector.fg_encoder.layer4.1.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1/__module.kp_detector.fg_encoder.layer4.1.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %278 : Tensor = aten::add(%out.31, %272, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %279 : Tensor = aten::relu(%278), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1/__module.kp_detector.fg_encoder.layer4.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 (CaskFlattenConvolution)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - CaskFlattenConvolution has no valid tactics for this config, skipping
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Half(4096,1:8,512,64), Float(32768,64,8,1) -> Float(32768,64,8,1) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %input.221 : Tensor = aten::_convolution(%275, %self.kp_detector.fg_encoder.layer4.1.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1/__module.kp_detector.fg_encoder.layer4.1.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.31 : Tensor = aten::batch_norm(%input.221, %self.kp_detector.fg_encoder.layer4.1.bn2.weight, %self.kp_detector.fg_encoder.layer4.1.bn2.bias, %self.kp_detector.fg_encoder.layer4.1.bn2.running_mean, %self.kp_detector.fg_encoder.layer4.1.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1/__module.kp_detector.fg_encoder.layer4.1.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %278 : Tensor = aten::add(%out.31, %272, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %279 : Tensor = aten::relu(%278), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1/__module.kp_detector.fg_encoder.layer4.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 (CaskConvolution)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - CaskConvolution has no valid tactics for this config, skipping
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %input.221 : Tensor = aten::_convolution(%275, %self.kp_detector.fg_encoder.layer4.1.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1/__module.kp_detector.fg_encoder.layer4.1.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.31 : Tensor = aten::batch_norm(%input.221, %self.kp_detector.fg_encoder.layer4.1.bn2.weight, %self.kp_detector.fg_encoder.layer4.1.bn2.bias, %self.kp_detector.fg_encoder.layer4.1.bn2.running_mean, %self.kp_detector.fg_encoder.layer4.1.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1/__module.kp_detector.fg_encoder.layer4.1.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %278 : Tensor = aten::add(%out.31, %272, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %279 : Tensor = aten::relu(%278), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1/__module.kp_detector.fg_encoder.layer4.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 (CaskFlattenConvolution)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - CaskFlattenConvolution has no valid tactics for this config, skipping
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Half(4096,1:8,512,64), Half(4096,1:8,512,64) -> Half(4096,1:8,512,64) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - =============== Computing costs for 
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Float(32768,64,8,1) -> Float(512,1,1,1) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Float(8192,1:4,1024,128) -> Float(128,1:4,128,128) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Half(32768,64,8,1) -> Half(512,1,1,1) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Half(16384,64:2,8,1) -> Half(256,1:2,1,1) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Half(4096,1:8,512,64) -> Half(64,1:8,64,64) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - =============== Computing costs for 
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Float(512,1,1,1) -> Float(100,1,1,1) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Unrecognized MMA instruction source format or shape: sm50_xmma_cublas_gemvx_f32f32_f32_f32_tn_n_int32_unit_n_launch_param64x2x1_strided_copyx_unit_stride
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Unrecognized MMA instruction source format or shape: sm50_xmma_cublas_gemvx_f32f32_f32_f32_tn_n_int32_unit_n_launch_param64x2x1_strided_copyx_unit_stride
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Unrecognized MMA instruction source format or shape: sm50_xmma_cublas_gemvx_f32f32_f32_f32_tn_n_int32_unit_n_launch_param64x2x1_strided_copyx_unit_stride
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Unrecognized MMA instruction source format or shape: sm50_xmma_cublas_gemvx_f32f32_f32_f32_tn_n_int32_unit_n_launch_param64x2x1_strided_copyx_unit_stride
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Unrecognized MMA instruction source format or shape: sm50_xmma_cublas_gemvx_f32f32_f32_f32_tn_n_int32_unit_n_launch_param64x2x1_strided_unit_stride
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Unrecognized MMA instruction source format or shape: sm50_xmma_cublas_gemvx_f32f32_f32_f32_tn_n_int32_unit_n_launch_param64x2x1_strided_unit_stride
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Unrecognized MMA instruction source format or shape: sm50_xmma_cublas_gemvx_f32f32_f32_f32_tn_n_int32_unit_n_launch_param64x2x1_strided_unit_stride
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Unrecognized MMA instruction source format or shape: sm50_xmma_cublas_gemvx_f32f32_f32_f32_tn_n_int32_unit_n_launch_param64x2x1_strided_unit_stride
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Unrecognized MMA instruction source format or shape: sm50_xmma_cublas_gemvx_f32f32_f32_f32_tn_n_int32_unit_n_launch_param4x32x32_strided_copyx_unit_stride
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Unrecognized MMA instruction source format or shape: sm50_xmma_cublas_gemvx_f32f32_f32_f32_tn_n_int32_unit_n_launch_param4x32x32_strided_copyx_unit_stride
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Unrecognized MMA instruction source format or shape: sm50_xmma_cublas_gemvx_f32f32_f32_f32_tn_n_int32_unit_n_launch_param4x32x32_strided_copyx_unit_stride
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Unrecognized MMA instruction source format or shape: sm50_xmma_cublas_gemvx_f32f32_f32_f32_tn_n_int32_unit_n_launch_param4x32x32_strided_copyx_unit_stride
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Unrecognized MMA instruction source format or shape: sm50_xmma_cublas_gemvx_f32f32_f32_f32_tn_n_int32_unit_n_launch_param16x32x32_strided_copyx_unit_stride
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Unrecognized MMA instruction source format or shape: sm50_xmma_cublas_gemvx_f32f32_f32_f32_tn_n_int32_unit_n_launch_param16x32x32_strided_copyx_unit_stride
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Unrecognized MMA instruction source format or shape: sm50_xmma_cublas_gemvx_f32f32_f32_f32_tn_n_int32_unit_n_launch_param16x32x32_strided_copyx_unit_stride
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Unrecognized MMA instruction source format or shape: sm50_xmma_cublas_gemvx_f32f32_f32_f32_tn_n_int32_unit_n_launch_param16x32x32_strided_copyx_unit_stride
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Unrecognized MMA instruction source format or shape: sm50_xmma_cublas_gemvx_f32f32_f32_f32_tn_n_int32_unit_n_launch_param4x32x32_strided_unit_stride
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Unrecognized MMA instruction source format or shape: sm50_xmma_cublas_gemvx_f32f32_f32_f32_tn_n_int32_unit_n_launch_param4x32x32_strided_unit_stride
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Unrecognized MMA instruction source format or shape: sm50_xmma_cublas_gemvx_f32f32_f32_f32_tn_n_int32_unit_n_launch_param4x32x32_strided_unit_stride
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Unrecognized MMA instruction source format or shape: sm50_xmma_cublas_gemvx_f32f32_f32_f32_tn_n_int32_unit_n_launch_param4x32x32_strided_unit_stride
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Unrecognized MMA instruction source format or shape: sm50_xmma_cublas_gemvx_f32f32_f32_f32_tn_n_int32_unit_n_launch_param8x32x32_strided_unit_stride
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Unrecognized MMA instruction source format or shape: sm50_xmma_cublas_gemvx_f32f32_f32_f32_tn_n_int32_unit_n_launch_param8x32x32_strided_unit_stride
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Unrecognized MMA instruction source format or shape: sm50_xmma_cublas_gemvx_f32f32_f32_f32_tn_n_int32_unit_n_launch_param8x32x32_strided_unit_stride
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Unrecognized MMA instruction source format or shape: sm50_xmma_cublas_gemvx_f32f32_f32_f32_tn_n_int32_unit_n_launch_param8x32x32_strided_unit_stride
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Unrecognized MMA instruction source format or shape: sm70_xmma_gemm_as_conv1x1_f32f32_f32_f32_tn_n_simt_small_batch_bias_relu
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Unrecognized MMA instruction source format or shape: sm50_xmma_cublas_gemvx_f32f32_f32_f32_tn_n_int32_unit_n_launch_param8x32x32_strided_copyx_unit_stride
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Unrecognized MMA instruction source format or shape: sm50_xmma_cublas_gemvx_f32f32_f32_f32_tn_n_int32_unit_n_launch_param8x32x32_strided_copyx_unit_stride
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Unrecognized MMA instruction source format or shape: sm50_xmma_cublas_gemvx_f32f32_f32_f32_tn_n_int32_unit_n_launch_param8x32x32_strided_copyx_unit_stride
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Unrecognized MMA instruction source format or shape: sm50_xmma_cublas_gemvx_f32f32_f32_f32_tn_n_int32_unit_n_launch_param8x32x32_strided_copyx_unit_stride
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Unrecognized MMA instruction source format or shape: sm50_xmma_cublas_gemvx_f32f32_f32_f32_tn_n_int32_unit_n_launch_param16x32x32_strided_unit_stride
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Unrecognized MMA instruction source format or shape: sm50_xmma_cublas_gemvx_f32f32_f32_f32_tn_n_int32_unit_n_launch_param16x32x32_strided_unit_stride
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Unrecognized MMA instruction source format or shape: sm50_xmma_cublas_gemvx_f32f32_f32_f32_tn_n_int32_unit_n_launch_param16x32x32_strided_unit_stride
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Unrecognized MMA instruction source format or shape: sm50_xmma_cublas_gemvx_f32f32_f32_f32_tn_n_int32_unit_n_launch_param16x32x32_strided_unit_stride
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - %362 : Tensor = aten::matmul(%input.341, %361) + [Freeze Tensor %363 : Tensor = trt::const(%self.kp_detector.fg_encoder.fc.bias) ] + (Unnamed Layer* 252) [Shuffle] + unsqueeze_node_after_[Freeze Tensor %363 : Tensor = trt::const(%self.kp_detector.fg_encoder.fc.bias) ] + (Unnamed Layer* 252) [Shuffle]_(Unnamed Layer* 252) [Shuffle]_output + %364 : Tensor = aten::add(%363, %362, %365) + PWN(%fg_kp.15 : Tensor = aten::sigmoid(%364), scope: __module.kp_detector # /home/joy/remote/stable-diffusion-backend/personalize/thin_plate_spline_motion_model/tpsnn.py:43:0): 33 available tactics, 33 unparsable, 0 pruned, 33 remaining after tactic pruning.
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Float(512,1,512,512) -> Float(100,1,100,100) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Float(128,1:4,128,128) -> Float(25,1:4,25,25) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - %362 : Tensor = aten::matmul(%input.341, %361) + [Freeze Tensor %363 : Tensor = trt::const(%self.kp_detector.fg_encoder.fc.bias) ] + (Unnamed Layer* 252) [Shuffle] + unsqueeze_node_after_[Freeze Tensor %363 : Tensor = trt::const(%self.kp_detector.fg_encoder.fc.bias) ] + (Unnamed Layer* 252) [Shuffle]_(Unnamed Layer* 252) [Shuffle]_output + %364 : Tensor = aten::add(%363, %362, %365) + PWN(%fg_kp.15 : Tensor = aten::sigmoid(%364), scope: __module.kp_detector # /home/joy/remote/stable-diffusion-backend/personalize/thin_plate_spline_motion_model/tpsnn.py:43:0): 14 available tactics, 2 unparsable, 6 pruned, 8 remaining after tactic pruning.
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Half(512,1,1,1) -> Half(100,1,1,1) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Half(256,1:2,1,1) -> Half(100,1,1,1) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %362 : Tensor = aten::matmul(%input.341, %361) + [Freeze Tensor %363 : Tensor = trt::const(%self.kp_detector.fg_encoder.fc.bias) ] + (Unnamed Layer* 252) [Shuffle] + unsqueeze_node_after_[Freeze Tensor %363 : Tensor = trt::const(%self.kp_detector.fg_encoder.fc.bias) ] + (Unnamed Layer* 252) [Shuffle]_(Unnamed Layer* 252) [Shuffle]_output + %364 : Tensor = aten::add(%363, %362, %365) + PWN(%fg_kp.15 : Tensor = aten::sigmoid(%364), scope: __module.kp_detector # /home/joy/remote/stable-diffusion-backend/personalize/thin_plate_spline_motion_model/tpsnn.py:43:0) (CaskConvolution)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - CaskConvolution has no valid tactics for this config, skipping
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %362 : Tensor = aten::matmul(%input.341, %361) + [Freeze Tensor %363 : Tensor = trt::const(%self.kp_detector.fg_encoder.fc.bias) ] + (Unnamed Layer* 252) [Shuffle] + unsqueeze_node_after_[Freeze Tensor %363 : Tensor = trt::const(%self.kp_detector.fg_encoder.fc.bias) ] + (Unnamed Layer* 252) [Shuffle]_(Unnamed Layer* 252) [Shuffle]_output + %364 : Tensor = aten::add(%363, %362, %365) + PWN(%fg_kp.15 : Tensor = aten::sigmoid(%364), scope: __module.kp_detector # /home/joy/remote/stable-diffusion-backend/personalize/thin_plate_spline_motion_model/tpsnn.py:43:0) (CaskFlattenConvolution)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - CaskFlattenConvolution has no valid tactics for this config, skipping
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Half(256,1:2,1,1) -> Half(50,1:2,1,1) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %362 : Tensor = aten::matmul(%input.341, %361) + [Freeze Tensor %363 : Tensor = trt::const(%self.kp_detector.fg_encoder.fc.bias) ] + (Unnamed Layer* 252) [Shuffle] + unsqueeze_node_after_[Freeze Tensor %363 : Tensor = trt::const(%self.kp_detector.fg_encoder.fc.bias) ] + (Unnamed Layer* 252) [Shuffle]_(Unnamed Layer* 252) [Shuffle]_output + %364 : Tensor = aten::add(%363, %362, %365) + PWN(%fg_kp.15 : Tensor = aten::sigmoid(%364), scope: __module.kp_detector # /home/joy/remote/stable-diffusion-backend/personalize/thin_plate_spline_motion_model/tpsnn.py:43:0) (FusedConvActConvolution)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - FusedConvActConvolution has no valid tactics for this config, skipping
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %362 : Tensor = aten::matmul(%input.341, %361) + [Freeze Tensor %363 : Tensor = trt::const(%self.kp_detector.fg_encoder.fc.bias) ] + (Unnamed Layer* 252) [Shuffle] + unsqueeze_node_after_[Freeze Tensor %363 : Tensor = trt::const(%self.kp_detector.fg_encoder.fc.bias) ] + (Unnamed Layer* 252) [Shuffle]_(Unnamed Layer* 252) [Shuffle]_output + %364 : Tensor = aten::add(%363, %362, %365) + PWN(%fg_kp.15 : Tensor = aten::sigmoid(%364), scope: __module.kp_detector # /home/joy/remote/stable-diffusion-backend/personalize/thin_plate_spline_motion_model/tpsnn.py:43:0) (CublasConvolution)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - CublasConvolution has no valid tactics for this config, skipping
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %362 : Tensor = aten::matmul(%input.341, %361) + [Freeze Tensor %363 : Tensor = trt::const(%self.kp_detector.fg_encoder.fc.bias) ] + (Unnamed Layer* 252) [Shuffle] + unsqueeze_node_after_[Freeze Tensor %363 : Tensor = trt::const(%self.kp_detector.fg_encoder.fc.bias) ] + (Unnamed Layer* 252) [Shuffle]_(Unnamed Layer* 252) [Shuffle]_output + %364 : Tensor = aten::add(%363, %362, %365) + PWN(%fg_kp.15 : Tensor = aten::sigmoid(%364), scope: __module.kp_detector # /home/joy/remote/stable-diffusion-backend/personalize/thin_plate_spline_motion_model/tpsnn.py:43:0) (CaskConvolution)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - CaskConvolution has no valid tactics for this config, skipping
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %362 : Tensor = aten::matmul(%input.341, %361) + [Freeze Tensor %363 : Tensor = trt::const(%self.kp_detector.fg_encoder.fc.bias) ] + (Unnamed Layer* 252) [Shuffle] + unsqueeze_node_after_[Freeze Tensor %363 : Tensor = trt::const(%self.kp_detector.fg_encoder.fc.bias) ] + (Unnamed Layer* 252) [Shuffle]_(Unnamed Layer* 252) [Shuffle]_output + %364 : Tensor = aten::add(%363, %362, %365) + PWN(%fg_kp.15 : Tensor = aten::sigmoid(%364), scope: __module.kp_detector # /home/joy/remote/stable-diffusion-backend/personalize/thin_plate_spline_motion_model/tpsnn.py:43:0) (CaskFlattenConvolution)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - CaskFlattenConvolution has no valid tactics for this config, skipping
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Half(128,1:4,128,128) -> Half(25,1:4,25,25) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %362 : Tensor = aten::matmul(%input.341, %361) + [Freeze Tensor %363 : Tensor = trt::const(%self.kp_detector.fg_encoder.fc.bias) ] + (Unnamed Layer* 252) [Shuffle] + unsqueeze_node_after_[Freeze Tensor %363 : Tensor = trt::const(%self.kp_detector.fg_encoder.fc.bias) ] + (Unnamed Layer* 252) [Shuffle]_(Unnamed Layer* 252) [Shuffle]_output + %364 : Tensor = aten::add(%363, %362, %365) + PWN(%fg_kp.15 : Tensor = aten::sigmoid(%364), scope: __module.kp_detector # /home/joy/remote/stable-diffusion-backend/personalize/thin_plate_spline_motion_model/tpsnn.py:43:0) (CublasConvolution)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - CublasConvolution has no valid tactics for this config, skipping
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %362 : Tensor = aten::matmul(%input.341, %361) + [Freeze Tensor %363 : Tensor = trt::const(%self.kp_detector.fg_encoder.fc.bias) ] + (Unnamed Layer* 252) [Shuffle] + unsqueeze_node_after_[Freeze Tensor %363 : Tensor = trt::const(%self.kp_detector.fg_encoder.fc.bias) ] + (Unnamed Layer* 252) [Shuffle]_(Unnamed Layer* 252) [Shuffle]_output + %364 : Tensor = aten::add(%363, %362, %365) + PWN(%fg_kp.15 : Tensor = aten::sigmoid(%364), scope: __module.kp_detector # /home/joy/remote/stable-diffusion-backend/personalize/thin_plate_spline_motion_model/tpsnn.py:43:0) (CaskConvolution)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - CaskConvolution has no valid tactics for this config, skipping
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %362 : Tensor = aten::matmul(%input.341, %361) + [Freeze Tensor %363 : Tensor = trt::const(%self.kp_detector.fg_encoder.fc.bias) ] + (Unnamed Layer* 252) [Shuffle] + unsqueeze_node_after_[Freeze Tensor %363 : Tensor = trt::const(%self.kp_detector.fg_encoder.fc.bias) ] + (Unnamed Layer* 252) [Shuffle]_(Unnamed Layer* 252) [Shuffle]_output + %364 : Tensor = aten::add(%363, %362, %365) + PWN(%fg_kp.15 : Tensor = aten::sigmoid(%364), scope: __module.kp_detector # /home/joy/remote/stable-diffusion-backend/personalize/thin_plate_spline_motion_model/tpsnn.py:43:0) (CaskFlattenConvolution)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - CaskFlattenConvolution has no valid tactics for this config, skipping
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Half(64,1:8,64,64) -> Float(100,1,1,1) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %362 : Tensor = aten::matmul(%input.341, %361) + [Freeze Tensor %363 : Tensor = trt::const(%self.kp_detector.fg_encoder.fc.bias) ] + (Unnamed Layer* 252) [Shuffle] + unsqueeze_node_after_[Freeze Tensor %363 : Tensor = trt::const(%self.kp_detector.fg_encoder.fc.bias) ] + (Unnamed Layer* 252) [Shuffle]_(Unnamed Layer* 252) [Shuffle]_output + %364 : Tensor = aten::add(%363, %362, %365) + PWN(%fg_kp.15 : Tensor = aten::sigmoid(%364), scope: __module.kp_detector # /home/joy/remote/stable-diffusion-backend/personalize/thin_plate_spline_motion_model/tpsnn.py:43:0) (CublasConvolution)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - CublasConvolution has no valid tactics for this config, skipping
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %362 : Tensor = aten::matmul(%input.341, %361) + [Freeze Tensor %363 : Tensor = trt::const(%self.kp_detector.fg_encoder.fc.bias) ] + (Unnamed Layer* 252) [Shuffle] + unsqueeze_node_after_[Freeze Tensor %363 : Tensor = trt::const(%self.kp_detector.fg_encoder.fc.bias) ] + (Unnamed Layer* 252) [Shuffle]_(Unnamed Layer* 252) [Shuffle]_output + %364 : Tensor = aten::add(%363, %362, %365) + PWN(%fg_kp.15 : Tensor = aten::sigmoid(%364), scope: __module.kp_detector # /home/joy/remote/stable-diffusion-backend/personalize/thin_plate_spline_motion_model/tpsnn.py:43:0) (CaskConvolution)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - CaskConvolution has no valid tactics for this config, skipping
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %362 : Tensor = aten::matmul(%input.341, %361) + [Freeze Tensor %363 : Tensor = trt::const(%self.kp_detector.fg_encoder.fc.bias) ] + (Unnamed Layer* 252) [Shuffle] + unsqueeze_node_after_[Freeze Tensor %363 : Tensor = trt::const(%self.kp_detector.fg_encoder.fc.bias) ] + (Unnamed Layer* 252) [Shuffle]_(Unnamed Layer* 252) [Shuffle]_output + %364 : Tensor = aten::add(%363, %362, %365) + PWN(%fg_kp.15 : Tensor = aten::sigmoid(%364), scope: __module.kp_detector # /home/joy/remote/stable-diffusion-backend/personalize/thin_plate_spline_motion_model/tpsnn.py:43:0) (CaskFlattenConvolution)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - CaskFlattenConvolution has no valid tactics for this config, skipping
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Half(64,1:8,64,64) -> Half(13,1:8,13,13) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Half(32,1:16,32,32) -> Half(7,1:16,7,7) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %362 : Tensor = aten::matmul(%input.341, %361) + [Freeze Tensor %363 : Tensor = trt::const(%self.kp_detector.fg_encoder.fc.bias) ] + (Unnamed Layer* 252) [Shuffle] + unsqueeze_node_after_[Freeze Tensor %363 : Tensor = trt::const(%self.kp_detector.fg_encoder.fc.bias) ] + (Unnamed Layer* 252) [Shuffle]_(Unnamed Layer* 252) [Shuffle]_output + %364 : Tensor = aten::add(%363, %362, %365) + PWN(%fg_kp.15 : Tensor = aten::sigmoid(%364), scope: __module.kp_detector # /home/joy/remote/stable-diffusion-backend/personalize/thin_plate_spline_motion_model/tpsnn.py:43:0) (CublasConvolution)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - CublasConvolution has no valid tactics for this config, skipping
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - =============== Computing costs for 
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Float(512,1,1,1) -> Float(100,1,1,1) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Unrecognized MMA instruction source format or shape: sm50_xmma_cublas_gemvx_f32f32_f32_f32_tn_n_int32_unit_n_launch_param64x2x1_strided_copyx_unit_stride
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Unrecognized MMA instruction source format or shape: sm50_xmma_cublas_gemvx_f32f32_f32_f32_tn_n_int32_unit_n_launch_param64x2x1_strided_copyx_unit_stride
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Unrecognized MMA instruction source format or shape: sm50_xmma_cublas_gemvx_f32f32_f32_f32_tn_n_int32_unit_n_launch_param64x2x1_strided_copyx_unit_stride
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Unrecognized MMA instruction source format or shape: sm50_xmma_cublas_gemvx_f32f32_f32_f32_tn_n_int32_unit_n_launch_param64x2x1_strided_copyx_unit_stride
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Unrecognized MMA instruction source format or shape: sm50_xmma_cublas_gemvx_f32f32_f32_f32_tn_n_int32_unit_n_launch_param64x2x1_strided_unit_stride
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Unrecognized MMA instruction source format or shape: sm50_xmma_cublas_gemvx_f32f32_f32_f32_tn_n_int32_unit_n_launch_param64x2x1_strided_unit_stride
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Unrecognized MMA instruction source format or shape: sm50_xmma_cublas_gemvx_f32f32_f32_f32_tn_n_int32_unit_n_launch_param64x2x1_strided_unit_stride
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Unrecognized MMA instruction source format or shape: sm50_xmma_cublas_gemvx_f32f32_f32_f32_tn_n_int32_unit_n_launch_param64x2x1_strided_unit_stride
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Unrecognized MMA instruction source format or shape: sm50_xmma_cublas_gemvx_f32f32_f32_f32_tn_n_int32_unit_n_launch_param4x32x32_strided_copyx_unit_stride
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Unrecognized MMA instruction source format or shape: sm50_xmma_cublas_gemvx_f32f32_f32_f32_tn_n_int32_unit_n_launch_param4x32x32_strided_copyx_unit_stride
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Unrecognized MMA instruction source format or shape: sm50_xmma_cublas_gemvx_f32f32_f32_f32_tn_n_int32_unit_n_launch_param4x32x32_strided_copyx_unit_stride
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Unrecognized MMA instruction source format or shape: sm50_xmma_cublas_gemvx_f32f32_f32_f32_tn_n_int32_unit_n_launch_param4x32x32_strided_copyx_unit_stride
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Unrecognized MMA instruction source format or shape: sm50_xmma_cublas_gemvx_f32f32_f32_f32_tn_n_int32_unit_n_launch_param16x32x32_strided_copyx_unit_stride
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Unrecognized MMA instruction source format or shape: sm50_xmma_cublas_gemvx_f32f32_f32_f32_tn_n_int32_unit_n_launch_param16x32x32_strided_copyx_unit_stride
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Unrecognized MMA instruction source format or shape: sm50_xmma_cublas_gemvx_f32f32_f32_f32_tn_n_int32_unit_n_launch_param16x32x32_strided_copyx_unit_stride
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Unrecognized MMA instruction source format or shape: sm50_xmma_cublas_gemvx_f32f32_f32_f32_tn_n_int32_unit_n_launch_param16x32x32_strided_copyx_unit_stride
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Unrecognized MMA instruction source format or shape: sm50_xmma_cublas_gemvx_f32f32_f32_f32_tn_n_int32_unit_n_launch_param4x32x32_strided_unit_stride
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Unrecognized MMA instruction source format or shape: sm50_xmma_cublas_gemvx_f32f32_f32_f32_tn_n_int32_unit_n_launch_param4x32x32_strided_unit_stride
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Unrecognized MMA instruction source format or shape: sm50_xmma_cublas_gemvx_f32f32_f32_f32_tn_n_int32_unit_n_launch_param4x32x32_strided_unit_stride
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Unrecognized MMA instruction source format or shape: sm50_xmma_cublas_gemvx_f32f32_f32_f32_tn_n_int32_unit_n_launch_param4x32x32_strided_unit_stride
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Unrecognized MMA instruction source format or shape: sm50_xmma_cublas_gemvx_f32f32_f32_f32_tn_n_int32_unit_n_launch_param8x32x32_strided_unit_stride
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Unrecognized MMA instruction source format or shape: sm50_xmma_cublas_gemvx_f32f32_f32_f32_tn_n_int32_unit_n_launch_param8x32x32_strided_unit_stride
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Unrecognized MMA instruction source format or shape: sm50_xmma_cublas_gemvx_f32f32_f32_f32_tn_n_int32_unit_n_launch_param8x32x32_strided_unit_stride
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Unrecognized MMA instruction source format or shape: sm50_xmma_cublas_gemvx_f32f32_f32_f32_tn_n_int32_unit_n_launch_param8x32x32_strided_unit_stride
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Unrecognized MMA instruction source format or shape: sm70_xmma_gemm_as_conv1x1_f32f32_f32_f32_tn_n_simt_small_batch_bias_relu
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Unrecognized MMA instruction source format or shape: sm50_xmma_cublas_gemvx_f32f32_f32_f32_tn_n_int32_unit_n_launch_param8x32x32_strided_copyx_unit_stride
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Unrecognized MMA instruction source format or shape: sm50_xmma_cublas_gemvx_f32f32_f32_f32_tn_n_int32_unit_n_launch_param8x32x32_strided_copyx_unit_stride
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Unrecognized MMA instruction source format or shape: sm50_xmma_cublas_gemvx_f32f32_f32_f32_tn_n_int32_unit_n_launch_param8x32x32_strided_copyx_unit_stride
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Unrecognized MMA instruction source format or shape: sm50_xmma_cublas_gemvx_f32f32_f32_f32_tn_n_int32_unit_n_launch_param8x32x32_strided_copyx_unit_stride
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Unrecognized MMA instruction source format or shape: sm50_xmma_cublas_gemvx_f32f32_f32_f32_tn_n_int32_unit_n_launch_param16x32x32_strided_unit_stride
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Unrecognized MMA instruction source format or shape: sm50_xmma_cublas_gemvx_f32f32_f32_f32_tn_n_int32_unit_n_launch_param16x32x32_strided_unit_stride
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Unrecognized MMA instruction source format or shape: sm50_xmma_cublas_gemvx_f32f32_f32_f32_tn_n_int32_unit_n_launch_param16x32x32_strided_unit_stride
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Unrecognized MMA instruction source format or shape: sm50_xmma_cublas_gemvx_f32f32_f32_f32_tn_n_int32_unit_n_launch_param16x32x32_strided_unit_stride
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - %283 : Tensor = aten::matmul(%input.227, %282) + [Freeze Tensor %284 : Tensor = trt::const(%self.kp_detector.fg_encoder.fc.bias) ] + (Unnamed Layer* 168) [Shuffle] + unsqueeze_node_after_[Freeze Tensor %284 : Tensor = trt::const(%self.kp_detector.fg_encoder.fc.bias) ] + (Unnamed Layer* 168) [Shuffle]_(Unnamed Layer* 168) [Shuffle]_output + %285 : Tensor = aten::add(%284, %283, %365) + PWN(%fg_kp.9 : Tensor = aten::sigmoid(%285), scope: __module.kp_detector # /home/joy/remote/stable-diffusion-backend/personalize/thin_plate_spline_motion_model/tpsnn.py:43:0): 33 available tactics, 33 unparsable, 0 pruned, 33 remaining after tactic pruning.
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Float(512,1,512,512) -> Float(100,1,100,100) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Float(128,1:4,128,128) -> Float(25,1:4,25,25) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - %283 : Tensor = aten::matmul(%input.227, %282) + [Freeze Tensor %284 : Tensor = trt::const(%self.kp_detector.fg_encoder.fc.bias) ] + (Unnamed Layer* 168) [Shuffle] + unsqueeze_node_after_[Freeze Tensor %284 : Tensor = trt::const(%self.kp_detector.fg_encoder.fc.bias) ] + (Unnamed Layer* 168) [Shuffle]_(Unnamed Layer* 168) [Shuffle]_output + %285 : Tensor = aten::add(%284, %283, %365) + PWN(%fg_kp.9 : Tensor = aten::sigmoid(%285), scope: __module.kp_detector # /home/joy/remote/stable-diffusion-backend/personalize/thin_plate_spline_motion_model/tpsnn.py:43:0): 14 available tactics, 2 unparsable, 6 pruned, 8 remaining after tactic pruning.
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Half(512,1,1,1) -> Half(100,1,1,1) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Half(256,1:2,1,1) -> Half(100,1,1,1) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %283 : Tensor = aten::matmul(%input.227, %282) + [Freeze Tensor %284 : Tensor = trt::const(%self.kp_detector.fg_encoder.fc.bias) ] + (Unnamed Layer* 168) [Shuffle] + unsqueeze_node_after_[Freeze Tensor %284 : Tensor = trt::const(%self.kp_detector.fg_encoder.fc.bias) ] + (Unnamed Layer* 168) [Shuffle]_(Unnamed Layer* 168) [Shuffle]_output + %285 : Tensor = aten::add(%284, %283, %365) + PWN(%fg_kp.9 : Tensor = aten::sigmoid(%285), scope: __module.kp_detector # /home/joy/remote/stable-diffusion-backend/personalize/thin_plate_spline_motion_model/tpsnn.py:43:0) (CaskConvolution)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - CaskConvolution has no valid tactics for this config, skipping
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %283 : Tensor = aten::matmul(%input.227, %282) + [Freeze Tensor %284 : Tensor = trt::const(%self.kp_detector.fg_encoder.fc.bias) ] + (Unnamed Layer* 168) [Shuffle] + unsqueeze_node_after_[Freeze Tensor %284 : Tensor = trt::const(%self.kp_detector.fg_encoder.fc.bias) ] + (Unnamed Layer* 168) [Shuffle]_(Unnamed Layer* 168) [Shuffle]_output + %285 : Tensor = aten::add(%284, %283, %365) + PWN(%fg_kp.9 : Tensor = aten::sigmoid(%285), scope: __module.kp_detector # /home/joy/remote/stable-diffusion-backend/personalize/thin_plate_spline_motion_model/tpsnn.py:43:0) (CaskFlattenConvolution)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - CaskFlattenConvolution has no valid tactics for this config, skipping
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Half(256,1:2,1,1) -> Half(50,1:2,1,1) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %283 : Tensor = aten::matmul(%input.227, %282) + [Freeze Tensor %284 : Tensor = trt::const(%self.kp_detector.fg_encoder.fc.bias) ] + (Unnamed Layer* 168) [Shuffle] + unsqueeze_node_after_[Freeze Tensor %284 : Tensor = trt::const(%self.kp_detector.fg_encoder.fc.bias) ] + (Unnamed Layer* 168) [Shuffle]_(Unnamed Layer* 168) [Shuffle]_output + %285 : Tensor = aten::add(%284, %283, %365) + PWN(%fg_kp.9 : Tensor = aten::sigmoid(%285), scope: __module.kp_detector # /home/joy/remote/stable-diffusion-backend/personalize/thin_plate_spline_motion_model/tpsnn.py:43:0) (FusedConvActConvolution)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - FusedConvActConvolution has no valid tactics for this config, skipping
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %283 : Tensor = aten::matmul(%input.227, %282) + [Freeze Tensor %284 : Tensor = trt::const(%self.kp_detector.fg_encoder.fc.bias) ] + (Unnamed Layer* 168) [Shuffle] + unsqueeze_node_after_[Freeze Tensor %284 : Tensor = trt::const(%self.kp_detector.fg_encoder.fc.bias) ] + (Unnamed Layer* 168) [Shuffle]_(Unnamed Layer* 168) [Shuffle]_output + %285 : Tensor = aten::add(%284, %283, %365) + PWN(%fg_kp.9 : Tensor = aten::sigmoid(%285), scope: __module.kp_detector # /home/joy/remote/stable-diffusion-backend/personalize/thin_plate_spline_motion_model/tpsnn.py:43:0) (CublasConvolution)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - CublasConvolution has no valid tactics for this config, skipping
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %283 : Tensor = aten::matmul(%input.227, %282) + [Freeze Tensor %284 : Tensor = trt::const(%self.kp_detector.fg_encoder.fc.bias) ] + (Unnamed Layer* 168) [Shuffle] + unsqueeze_node_after_[Freeze Tensor %284 : Tensor = trt::const(%self.kp_detector.fg_encoder.fc.bias) ] + (Unnamed Layer* 168) [Shuffle]_(Unnamed Layer* 168) [Shuffle]_output + %285 : Tensor = aten::add(%284, %283, %365) + PWN(%fg_kp.9 : Tensor = aten::sigmoid(%285), scope: __module.kp_detector # /home/joy/remote/stable-diffusion-backend/personalize/thin_plate_spline_motion_model/tpsnn.py:43:0) (CaskConvolution)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - CaskConvolution has no valid tactics for this config, skipping
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %283 : Tensor = aten::matmul(%input.227, %282) + [Freeze Tensor %284 : Tensor = trt::const(%self.kp_detector.fg_encoder.fc.bias) ] + (Unnamed Layer* 168) [Shuffle] + unsqueeze_node_after_[Freeze Tensor %284 : Tensor = trt::const(%self.kp_detector.fg_encoder.fc.bias) ] + (Unnamed Layer* 168) [Shuffle]_(Unnamed Layer* 168) [Shuffle]_output + %285 : Tensor = aten::add(%284, %283, %365) + PWN(%fg_kp.9 : Tensor = aten::sigmoid(%285), scope: __module.kp_detector # /home/joy/remote/stable-diffusion-backend/personalize/thin_plate_spline_motion_model/tpsnn.py:43:0) (CaskFlattenConvolution)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - CaskFlattenConvolution has no valid tactics for this config, skipping
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Half(128,1:4,128,128) -> Half(25,1:4,25,25) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %283 : Tensor = aten::matmul(%input.227, %282) + [Freeze Tensor %284 : Tensor = trt::const(%self.kp_detector.fg_encoder.fc.bias) ] + (Unnamed Layer* 168) [Shuffle] + unsqueeze_node_after_[Freeze Tensor %284 : Tensor = trt::const(%self.kp_detector.fg_encoder.fc.bias) ] + (Unnamed Layer* 168) [Shuffle]_(Unnamed Layer* 168) [Shuffle]_output + %285 : Tensor = aten::add(%284, %283, %365) + PWN(%fg_kp.9 : Tensor = aten::sigmoid(%285), scope: __module.kp_detector # /home/joy/remote/stable-diffusion-backend/personalize/thin_plate_spline_motion_model/tpsnn.py:43:0) (CublasConvolution)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - CublasConvolution has no valid tactics for this config, skipping
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %283 : Tensor = aten::matmul(%input.227, %282) + [Freeze Tensor %284 : Tensor = trt::const(%self.kp_detector.fg_encoder.fc.bias) ] + (Unnamed Layer* 168) [Shuffle] + unsqueeze_node_after_[Freeze Tensor %284 : Tensor = trt::const(%self.kp_detector.fg_encoder.fc.bias) ] + (Unnamed Layer* 168) [Shuffle]_(Unnamed Layer* 168) [Shuffle]_output + %285 : Tensor = aten::add(%284, %283, %365) + PWN(%fg_kp.9 : Tensor = aten::sigmoid(%285), scope: __module.kp_detector # /home/joy/remote/stable-diffusion-backend/personalize/thin_plate_spline_motion_model/tpsnn.py:43:0) (CaskConvolution)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - CaskConvolution has no valid tactics for this config, skipping
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %283 : Tensor = aten::matmul(%input.227, %282) + [Freeze Tensor %284 : Tensor = trt::const(%self.kp_detector.fg_encoder.fc.bias) ] + (Unnamed Layer* 168) [Shuffle] + unsqueeze_node_after_[Freeze Tensor %284 : Tensor = trt::const(%self.kp_detector.fg_encoder.fc.bias) ] + (Unnamed Layer* 168) [Shuffle]_(Unnamed Layer* 168) [Shuffle]_output + %285 : Tensor = aten::add(%284, %283, %365) + PWN(%fg_kp.9 : Tensor = aten::sigmoid(%285), scope: __module.kp_detector # /home/joy/remote/stable-diffusion-backend/personalize/thin_plate_spline_motion_model/tpsnn.py:43:0) (CaskFlattenConvolution)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - CaskFlattenConvolution has no valid tactics for this config, skipping
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Half(64,1:8,64,64) -> Float(100,1,1,1) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %283 : Tensor = aten::matmul(%input.227, %282) + [Freeze Tensor %284 : Tensor = trt::const(%self.kp_detector.fg_encoder.fc.bias) ] + (Unnamed Layer* 168) [Shuffle] + unsqueeze_node_after_[Freeze Tensor %284 : Tensor = trt::const(%self.kp_detector.fg_encoder.fc.bias) ] + (Unnamed Layer* 168) [Shuffle]_(Unnamed Layer* 168) [Shuffle]_output + %285 : Tensor = aten::add(%284, %283, %365) + PWN(%fg_kp.9 : Tensor = aten::sigmoid(%285), scope: __module.kp_detector # /home/joy/remote/stable-diffusion-backend/personalize/thin_plate_spline_motion_model/tpsnn.py:43:0) (CublasConvolution)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - CublasConvolution has no valid tactics for this config, skipping
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %283 : Tensor = aten::matmul(%input.227, %282) + [Freeze Tensor %284 : Tensor = trt::const(%self.kp_detector.fg_encoder.fc.bias) ] + (Unnamed Layer* 168) [Shuffle] + unsqueeze_node_after_[Freeze Tensor %284 : Tensor = trt::const(%self.kp_detector.fg_encoder.fc.bias) ] + (Unnamed Layer* 168) [Shuffle]_(Unnamed Layer* 168) [Shuffle]_output + %285 : Tensor = aten::add(%284, %283, %365) + PWN(%fg_kp.9 : Tensor = aten::sigmoid(%285), scope: __module.kp_detector # /home/joy/remote/stable-diffusion-backend/personalize/thin_plate_spline_motion_model/tpsnn.py:43:0) (CaskConvolution)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - CaskConvolution has no valid tactics for this config, skipping
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %283 : Tensor = aten::matmul(%input.227, %282) + [Freeze Tensor %284 : Tensor = trt::const(%self.kp_detector.fg_encoder.fc.bias) ] + (Unnamed Layer* 168) [Shuffle] + unsqueeze_node_after_[Freeze Tensor %284 : Tensor = trt::const(%self.kp_detector.fg_encoder.fc.bias) ] + (Unnamed Layer* 168) [Shuffle]_(Unnamed Layer* 168) [Shuffle]_output + %285 : Tensor = aten::add(%284, %283, %365) + PWN(%fg_kp.9 : Tensor = aten::sigmoid(%285), scope: __module.kp_detector # /home/joy/remote/stable-diffusion-backend/personalize/thin_plate_spline_motion_model/tpsnn.py:43:0) (CaskFlattenConvolution)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - CaskFlattenConvolution has no valid tactics for this config, skipping
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Half(64,1:8,64,64) -> Half(13,1:8,13,13) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Half(32,1:16,32,32) -> Half(7,1:16,7,7) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %283 : Tensor = aten::matmul(%input.227, %282) + [Freeze Tensor %284 : Tensor = trt::const(%self.kp_detector.fg_encoder.fc.bias) ] + (Unnamed Layer* 168) [Shuffle] + unsqueeze_node_after_[Freeze Tensor %284 : Tensor = trt::const(%self.kp_detector.fg_encoder.fc.bias) ] + (Unnamed Layer* 168) [Shuffle]_(Unnamed Layer* 168) [Shuffle]_output + %285 : Tensor = aten::add(%284, %283, %365) + PWN(%fg_kp.9 : Tensor = aten::sigmoid(%285), scope: __module.kp_detector # /home/joy/remote/stable-diffusion-backend/personalize/thin_plate_spline_motion_model/tpsnn.py:43:0) (CublasConvolution)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - CublasConvolution has no valid tactics for this config, skipping
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - =============== Computing costs for 
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Float(100,1,1,1), Float(1,1,1,1) -> Float(100,1,1,1) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %368 : Tensor = aten::mul(%fg_kp.15, %206), scope: __module.kp_detector # /home/joy/remote/stable-diffusion-backend/personalize/thin_plate_spline_motion_model/tpsnn.py:43:0 (ElementWise)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Tactic: 0x0000000000000001 Time: 0.00248043
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Fastest Tactic: 0x0000000000000001 Time: 0.00248043
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - >>>>>>>>>>>>>>> Chose Runner Type: ElementWise Tactic: 0x0000000000000001
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Float(25,1:4,25,25), Float(1,1:4,1,1) -> Float(25,1:4,25,25) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %368 : Tensor = aten::mul(%fg_kp.15, %206), scope: __module.kp_detector # /home/joy/remote/stable-diffusion-backend/personalize/thin_plate_spline_motion_model/tpsnn.py:43:0 (ElementWise)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Tactic: 0x0000000000000001 Time: 0.0024401
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Tactic: 0x0000000000000002 Time: 0.00315794
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Fastest Tactic: 0x0000000000000001 Time: 0.0024401
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - >>>>>>>>>>>>>>> Chose Runner Type: ElementWise Tactic: 0x0000000000000001
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Float(4,1:32,1,1), Float(1,1:32,1,1) -> Float(4,1:32,1,1) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %368 : Tensor = aten::mul(%fg_kp.15, %206), scope: __module.kp_detector # /home/joy/remote/stable-diffusion-backend/personalize/thin_plate_spline_motion_model/tpsnn.py:43:0 (ElementWise)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Tactic: 0x0000000000000001 Time: 0.00257649
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Fastest Tactic: 0x0000000000000001 Time: 0.00257649
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - >>>>>>>>>>>>>>> Chose Runner Type: ElementWise Tactic: 0x0000000000000001
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Half(100,1,1,1), Half(1,1,1,1) -> Half(100,1,1,1) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %368 : Tensor = aten::mul(%fg_kp.15, %206), scope: __module.kp_detector # /home/joy/remote/stable-diffusion-backend/personalize/thin_plate_spline_motion_model/tpsnn.py:43:0 (ElementWise)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Tactic: 0x0000000000000001 Time: 0.00273889
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Fastest Tactic: 0x0000000000000001 Time: 0.00273889
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - >>>>>>>>>>>>>>> Chose Runner Type: ElementWise Tactic: 0x0000000000000001
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Half(50,1:2,1,1), Half(1,1:2,1,1) -> Half(50,1:2,1,1) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %368 : Tensor = aten::mul(%fg_kp.15, %206), scope: __module.kp_detector # /home/joy/remote/stable-diffusion-backend/personalize/thin_plate_spline_motion_model/tpsnn.py:43:0 (ElementWise)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Tactic: 0x0000000000000001 Time: 0.00275658
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Fastest Tactic: 0x0000000000000001 Time: 0.00275658
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - >>>>>>>>>>>>>>> Chose Runner Type: ElementWise Tactic: 0x0000000000000001
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Half(13,1:8,13,13), Half(1,1:8,1,1) -> Half(13,1:8,13,13) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %368 : Tensor = aten::mul(%fg_kp.15, %206), scope: __module.kp_detector # /home/joy/remote/stable-diffusion-backend/personalize/thin_plate_spline_motion_model/tpsnn.py:43:0 (ElementWise)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Tactic: 0x0000000000000001 Time: 0.00288956
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Fastest Tactic: 0x0000000000000001 Time: 0.00288956
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - >>>>>>>>>>>>>>> Chose Runner Type: ElementWise Tactic: 0x0000000000000001
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - =============== Computing costs for 
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Float(100,1,1,1) -> Float(100,1) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Half(100,1,1,1) -> Half(100,1) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - =============== Computing costs for 
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Float(100,1), Float(1,1) -> Float(100,1) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %fg_kp : Tensor = aten::sub(%368, %208, %365), scope: __module.kp_detector # /home/joy/remote/stable-diffusion-backend/personalize/thin_plate_spline_motion_model/tpsnn.py:43:0 (ElementWise)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Tactic: 0x0000000000000001 Time: 0.00244866
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Fastest Tactic: 0x0000000000000001 Time: 0.00244866
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - >>>>>>>>>>>>>>> Chose Runner Type: ElementWise Tactic: 0x0000000000000001
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Half(100,1), Half(1,1) -> Half(100,1) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %fg_kp : Tensor = aten::sub(%368, %208, %365), scope: __module.kp_detector # /home/joy/remote/stable-diffusion-backend/personalize/thin_plate_spline_motion_model/tpsnn.py:43:0 (ElementWise)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Tactic: 0x0000000000000001 Time: 0.00273993
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Fastest Tactic: 0x0000000000000001 Time: 0.00273993
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - >>>>>>>>>>>>>>> Chose Runner Type: ElementWise Tactic: 0x0000000000000001
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - =============== Computing costs for 
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Float(100,1) -> Float(100,2,1) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Half(100,1) -> Half(100,2,1) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - =============== Computing costs for 
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Float(100,1,1,1), Float(1,1,1,1) -> Float(100,1,1,1) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %289 : Tensor = aten::mul(%fg_kp.9, %206), scope: __module.kp_detector # /home/joy/remote/stable-diffusion-backend/personalize/thin_plate_spline_motion_model/tpsnn.py:43:0 (ElementWise)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Tactic: 0x0000000000000001 Time: 0.00244679
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Fastest Tactic: 0x0000000000000001 Time: 0.00244679
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - >>>>>>>>>>>>>>> Chose Runner Type: ElementWise Tactic: 0x0000000000000001
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Float(25,1:4,25,25), Float(1,1:4,1,1) -> Float(25,1:4,25,25) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %289 : Tensor = aten::mul(%fg_kp.9, %206), scope: __module.kp_detector # /home/joy/remote/stable-diffusion-backend/personalize/thin_plate_spline_motion_model/tpsnn.py:43:0 (ElementWise)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Tactic: 0x0000000000000001 Time: 0.00244267
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Tactic: 0x0000000000000002 Time: 0.00315874
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Fastest Tactic: 0x0000000000000001 Time: 0.00244267
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - >>>>>>>>>>>>>>> Chose Runner Type: ElementWise Tactic: 0x0000000000000001
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Float(4,1:32,1,1), Float(1,1:32,1,1) -> Float(4,1:32,1,1) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %289 : Tensor = aten::mul(%fg_kp.9, %206), scope: __module.kp_detector # /home/joy/remote/stable-diffusion-backend/personalize/thin_plate_spline_motion_model/tpsnn.py:43:0 (ElementWise)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Tactic: 0x0000000000000001 Time: 0.00262694
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Fastest Tactic: 0x0000000000000001 Time: 0.00262694
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - >>>>>>>>>>>>>>> Chose Runner Type: ElementWise Tactic: 0x0000000000000001
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Half(100,1,1,1), Half(1,1,1,1) -> Half(100,1,1,1) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %289 : Tensor = aten::mul(%fg_kp.9, %206), scope: __module.kp_detector # /home/joy/remote/stable-diffusion-backend/personalize/thin_plate_spline_motion_model/tpsnn.py:43:0 (ElementWise)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Tactic: 0x0000000000000001 Time: 0.0027534
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Fastest Tactic: 0x0000000000000001 Time: 0.0027534
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - >>>>>>>>>>>>>>> Chose Runner Type: ElementWise Tactic: 0x0000000000000001
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Half(50,1:2,1,1), Half(1,1:2,1,1) -> Half(50,1:2,1,1) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %289 : Tensor = aten::mul(%fg_kp.9, %206), scope: __module.kp_detector # /home/joy/remote/stable-diffusion-backend/personalize/thin_plate_spline_motion_model/tpsnn.py:43:0 (ElementWise)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Tactic: 0x0000000000000001 Time: 0.00275086
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Fastest Tactic: 0x0000000000000001 Time: 0.00275086
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - >>>>>>>>>>>>>>> Chose Runner Type: ElementWise Tactic: 0x0000000000000001
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Half(13,1:8,13,13), Half(1,1:8,1,1) -> Half(13,1:8,13,13) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %289 : Tensor = aten::mul(%fg_kp.9, %206), scope: __module.kp_detector # /home/joy/remote/stable-diffusion-backend/personalize/thin_plate_spline_motion_model/tpsnn.py:43:0 (ElementWise)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Tactic: 0x0000000000000001 Time: 0.00286806
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Fastest Tactic: 0x0000000000000001 Time: 0.00286806
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - >>>>>>>>>>>>>>> Chose Runner Type: ElementWise Tactic: 0x0000000000000001
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - =============== Computing costs for 
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Float(100,1,1,1) -> Float(100,1) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Half(100,1,1,1) -> Half(100,1) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - =============== Computing costs for 
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Float(100,1), Float(1,1) -> Float(100,1) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %fg_kp.11 : Tensor = aten::sub(%289, %208, %365), scope: __module.kp_detector # /home/joy/remote/stable-diffusion-backend/personalize/thin_plate_spline_motion_model/tpsnn.py:43:0 (ElementWise)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Tactic: 0x0000000000000001 Time: 0.00245796
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Fastest Tactic: 0x0000000000000001 Time: 0.00245796
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - >>>>>>>>>>>>>>> Chose Runner Type: ElementWise Tactic: 0x0000000000000001
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Half(100,1), Half(1,1) -> Half(100,1) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %fg_kp.11 : Tensor = aten::sub(%289, %208, %365), scope: __module.kp_detector # /home/joy/remote/stable-diffusion-backend/personalize/thin_plate_spline_motion_model/tpsnn.py:43:0 (ElementWise)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Tactic: 0x0000000000000001 Time: 0.00274326
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Fastest Tactic: 0x0000000000000001 Time: 0.00274326
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - >>>>>>>>>>>>>>> Chose Runner Type: ElementWise Tactic: 0x0000000000000001
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - =============== Computing costs for 
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Float(100,1) -> Float(100,2,1) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Half(100,1) -> Half(100,2,1) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - =============== Computing costs for 
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Float(100,2,1) -> Float(2,1) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Half(100,2,1) -> Half(2,1) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - =============== Computing costs for 
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Float(2,1), Float(2,1), Float(1,1) -> Float(2,1) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: PWN(%mean_diff : Tensor = aten::sub(%372, %374, %365) # /home/joy/remote/stable-diffusion-backend/personalize/thin_plate_spline_motion_model/tpsnn.py:503:0, %376 : Tensor = aten::pow(%mean_diff, %21) # /home/joy/remote/stable-diffusion-backend/personalize/thin_plate_spline_motion_model/tpsnn.py:503:0) (PointWiseV2)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Tactic: 0x0000000000000000 Time: 0.00256624
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Tactic: 0x0000000000000001 Time: 0.00264399
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Tactic: 0x0000000000000002 Time: 0.00257206
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Tactic: 0x0000000000000003 Time: 0.00266778
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Tactic: 0x0000000000000004 Time: 0.00276981
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Tactic: 0x0000000000000005 Time: 0.00256456
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Tactic: 0x0000000000000006 Time: 0.00287116
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Tactic: 0x0000000000000007 Time: 0.00303835
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Tactic: 0x0000000000000008 Time: 0.00290811
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Tactic: 0x0000000000000009 Time: 0.00288653
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Tactic: 0x000000000000001c Time: 0.00276267
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Fastest Tactic: 0x0000000000000005 Time: 0.00256456
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: PWN(%mean_diff : Tensor = aten::sub(%372, %374, %365) # /home/joy/remote/stable-diffusion-backend/personalize/thin_plate_spline_motion_model/tpsnn.py:503:0, %376 : Tensor = aten::pow(%mean_diff, %21) # /home/joy/remote/stable-diffusion-backend/personalize/thin_plate_spline_motion_model/tpsnn.py:503:0) (PointWise)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - PointWise has no valid tactics for this config, skipping
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - >>>>>>>>>>>>>>> Chose Runner Type: PointWiseV2 Tactic: 0x0000000000000005
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Half(2,1), Half(2,1), Half(1,1) -> Half(2,1) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: PWN(%mean_diff : Tensor = aten::sub(%372, %374, %365) # /home/joy/remote/stable-diffusion-backend/personalize/thin_plate_spline_motion_model/tpsnn.py:503:0, %376 : Tensor = aten::pow(%mean_diff, %21) # /home/joy/remote/stable-diffusion-backend/personalize/thin_plate_spline_motion_model/tpsnn.py:503:0) (PointWiseV2)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Tactic: 0x0000000000000000 Time: 0.00275641
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Tactic: 0x0000000000000001 Time: 0.00319563
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Tactic: 0x0000000000000002 Time: 0.00270357
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Tactic: 0x0000000000000003 Time: 0.00288202
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Tactic: 0x0000000000000004 Time: 0.00293811
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Tactic: 0x0000000000000005 Time: 0.00269875
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Tactic: 0x0000000000000006 Time: 0.00300973
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Tactic: 0x0000000000000007 Time: 0.00307396
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Tactic: 0x0000000000000008 Time: 0.002931
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Tactic: 0x0000000000000009 Time: 0.00284203
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Tactic: 0x000000000000001c Time: 0.00272676
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Fastest Tactic: 0x0000000000000005 Time: 0.00269875
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: PWN(%mean_diff : Tensor = aten::sub(%372, %374, %365) # /home/joy/remote/stable-diffusion-backend/personalize/thin_plate_spline_motion_model/tpsnn.py:503:0, %376 : Tensor = aten::pow(%mean_diff, %21) # /home/joy/remote/stable-diffusion-backend/personalize/thin_plate_spline_motion_model/tpsnn.py:503:0) (PointWise)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - PointWise has no valid tactics for this config, skipping
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - >>>>>>>>>>>>>>> Chose Runner Type: PointWiseV2 Tactic: 0x0000000000000005
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - =============== Computing costs for 
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Float(2,1) -> Float() ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %377 : Tensor = aten::sum(%376, %5) # /home/joy/remote/stable-diffusion-backend/personalize/thin_plate_spline_motion_model/tpsnn.py:503:0 (Reduce)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Tactic: 0x0000000000000001 Time: 0.00272702
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Tactic: 0x0000000000000003 Time: 0.00273556
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Tactic: 0x0000000000000004 Time: 0.0047055
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Tactic: 0x0000000000000007 Time: 0.00272512
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Tactic: 0x0000000000000008 Time: 0.0047907
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Fastest Tactic: 0x0000000000000007 Time: 0.00272512
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - >>>>>>>>>>>>>>> Chose Runner Type: Reduce Tactic: 0x0000000000000007
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Half(2,1) -> Half() ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %377 : Tensor = aten::sum(%376, %5) # /home/joy/remote/stable-diffusion-backend/personalize/thin_plate_spline_motion_model/tpsnn.py:503:0 (Reduce)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Tactic: 0x0000000000000001 Time: 0.00273399
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Tactic: 0x0000000000000003 Time: 0.00274133
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Tactic: 0x0000000000000004 Time: 0.00477943
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Tactic: 0x0000000000000007 Time: 0.00273434
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Tactic: 0x0000000000000008 Time: 0.00479421
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Fastest Tactic: 0x0000000000000001 Time: 0.00273399
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - >>>>>>>>>>>>>>> Chose Runner Type: Reduce Tactic: 0x0000000000000001
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - =============== Computing costs for 
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Float(), Float() -> Float() ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %adapt_movement_scale : Tensor = aten::mul(%377, %206) # /home/joy/remote/stable-diffusion-backend/personalize/thin_plate_spline_motion_model/tpsnn.py:503:0 (ElementWise)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Tactic: 0x0000000000000001 Time: 0.0026055
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Fastest Tactic: 0x0000000000000001 Time: 0.0026055
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - >>>>>>>>>>>>>>> Chose Runner Type: ElementWise Tactic: 0x0000000000000001
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Half(), Half() -> Half() ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: %adapt_movement_scale : Tensor = aten::mul(%377, %206) # /home/joy/remote/stable-diffusion-backend/personalize/thin_plate_spline_motion_model/tpsnn.py:503:0 (ElementWise)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Tactic: 0x0000000000000001 Time: 0.00257067
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Fastest Tactic: 0x0000000000000001 Time: 0.00257067
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - >>>>>>>>>>>>>>> Chose Runner Type: ElementWise Tactic: 0x0000000000000001
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - =============== Computing costs for 
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Float() -> Float(1,1,1) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: (Unnamed Layer* 276) [Shuffle] (Shuffle)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Tactic: 0x0000000000000000 Time: 0.00239831
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Fastest Tactic: 0x0000000000000000 Time: 0.00239831
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - >>>>>>>>>>>>>>> Chose Runner Type: Shuffle Tactic: 0x0000000000000000
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Half() -> Half(1,1,1) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: (Unnamed Layer* 276) [Shuffle] (Shuffle)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Tactic: 0x0000000000000000 Time: 0.00250245
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Fastest Tactic: 0x0000000000000000 Time: 0.00250245
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - >>>>>>>>>>>>>>> Chose Runner Type: Shuffle Tactic: 0x0000000000000000
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - =============== Computing costs for 
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Float(100,2,1), Float(100,2,1), Float(1,1,1), Float(100,2,1) -> Float(100,2,1) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: PWN(PWN(%kp_value_diff.1 : Tensor = aten::sub(%371, %292, %365) # /home/joy/remote/stable-diffusion-backend/personalize/thin_plate_spline_motion_model/tpsnn.py:510:0, %380 : Tensor = aten::mul(%kp_value_diff.1, %adapt_movement_scale) # /home/joy/remote/stable-diffusion-backend/personalize/thin_plate_spline_motion_model/tpsnn.py:511:0), %kp.1 : Tensor = aten::add(%380, %211, %365) # /home/joy/remote/stable-diffusion-backend/personalize/thin_plate_spline_motion_model/tpsnn.py:511:0) (PointWiseV2)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Tactic: 0x0000000000000000 Time: 0.00271766
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Tactic: 0x0000000000000001 Time: 0.00291775
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Tactic: 0x0000000000000002 Time: 0.0027009
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Tactic: 0x0000000000000003 Time: 0.00303302
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Tactic: 0x0000000000000004 Time: 0.00358664
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Tactic: 0x0000000000000005 Time: 0.00289864
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Tactic: 0x0000000000000006 Time: 0.00316307
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Tactic: 0x0000000000000007 Time: 0.00319736
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Tactic: 0x0000000000000008 Time: 0.00302866
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Tactic: 0x0000000000000009 Time: 0.00290708
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Tactic: 0x000000000000001c Time: 0.00270908
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Fastest Tactic: 0x0000000000000002 Time: 0.0027009
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: PWN(PWN(%kp_value_diff.1 : Tensor = aten::sub(%371, %292, %365) # /home/joy/remote/stable-diffusion-backend/personalize/thin_plate_spline_motion_model/tpsnn.py:510:0, %380 : Tensor = aten::mul(%kp_value_diff.1, %adapt_movement_scale) # /home/joy/remote/stable-diffusion-backend/personalize/thin_plate_spline_motion_model/tpsnn.py:511:0), %kp.1 : Tensor = aten::add(%380, %211, %365) # /home/joy/remote/stable-diffusion-backend/personalize/thin_plate_spline_motion_model/tpsnn.py:511:0) (PointWise)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - PointWise has no valid tactics for this config, skipping
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - >>>>>>>>>>>>>>> Chose Runner Type: PointWiseV2 Tactic: 0x0000000000000002
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Float(1,2,1), Float(1,2,1), Float(1,1,1), Float(1,2,1) -> Float(1,2,1) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: PWN(PWN(%kp_value_diff.1 : Tensor = aten::sub(%371, %292, %365) # /home/joy/remote/stable-diffusion-backend/personalize/thin_plate_spline_motion_model/tpsnn.py:510:0, %380 : Tensor = aten::mul(%kp_value_diff.1, %adapt_movement_scale) # /home/joy/remote/stable-diffusion-backend/personalize/thin_plate_spline_motion_model/tpsnn.py:511:0), %kp.1 : Tensor = aten::add(%380, %211, %365) # /home/joy/remote/stable-diffusion-backend/personalize/thin_plate_spline_motion_model/tpsnn.py:511:0) (PointWiseV2)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - PointWiseV2 has no valid tactics for this config, skipping
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: PWN(PWN(%kp_value_diff.1 : Tensor = aten::sub(%371, %292, %365) # /home/joy/remote/stable-diffusion-backend/personalize/thin_plate_spline_motion_model/tpsnn.py:510:0, %380 : Tensor = aten::mul(%kp_value_diff.1, %adapt_movement_scale) # /home/joy/remote/stable-diffusion-backend/personalize/thin_plate_spline_motion_model/tpsnn.py:511:0), %kp.1 : Tensor = aten::add(%380, %211, %365) # /home/joy/remote/stable-diffusion-backend/personalize/thin_plate_spline_motion_model/tpsnn.py:511:0) (PointWise)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - PointWise has no valid tactics for this config, skipping
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Float(1:4,2,1), Float(1:4,2,1), Float(1:4,1,1), Float(1:4,2,1) -> Float(1:4,2,1) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: PWN(PWN(%kp_value_diff.1 : Tensor = aten::sub(%371, %292, %365) # /home/joy/remote/stable-diffusion-backend/personalize/thin_plate_spline_motion_model/tpsnn.py:510:0, %380 : Tensor = aten::mul(%kp_value_diff.1, %adapt_movement_scale) # /home/joy/remote/stable-diffusion-backend/personalize/thin_plate_spline_motion_model/tpsnn.py:511:0), %kp.1 : Tensor = aten::add(%380, %211, %365) # /home/joy/remote/stable-diffusion-backend/personalize/thin_plate_spline_motion_model/tpsnn.py:511:0) (PointWiseV2)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Tactic: 0x000000000000001c Time: 0.00317364
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Tactic: 0x000000000000001d Time: 0.0031938
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Tactic: 0x000000000000001e Time: 0.00290616
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Fastest Tactic: 0x000000000000001e Time: 0.00290616
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: PWN(PWN(%kp_value_diff.1 : Tensor = aten::sub(%371, %292, %365) # /home/joy/remote/stable-diffusion-backend/personalize/thin_plate_spline_motion_model/tpsnn.py:510:0, %380 : Tensor = aten::mul(%kp_value_diff.1, %adapt_movement_scale) # /home/joy/remote/stable-diffusion-backend/personalize/thin_plate_spline_motion_model/tpsnn.py:511:0), %kp.1 : Tensor = aten::add(%380, %211, %365) # /home/joy/remote/stable-diffusion-backend/personalize/thin_plate_spline_motion_model/tpsnn.py:511:0) (PointWise)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - PointWise has no valid tactics for this config, skipping
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - >>>>>>>>>>>>>>> Chose Runner Type: PointWiseV2 Tactic: 0x000000000000001e
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Float(100:32,2,1), Float(100:32,2,1), Float(1:32,1,1), Float(100:32,2,1) -> Float(100:32,2,1) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: PWN(PWN(%kp_value_diff.1 : Tensor = aten::sub(%371, %292, %365) # /home/joy/remote/stable-diffusion-backend/personalize/thin_plate_spline_motion_model/tpsnn.py:510:0, %380 : Tensor = aten::mul(%kp_value_diff.1, %adapt_movement_scale) # /home/joy/remote/stable-diffusion-backend/personalize/thin_plate_spline_motion_model/tpsnn.py:511:0), %kp.1 : Tensor = aten::add(%380, %211, %365) # /home/joy/remote/stable-diffusion-backend/personalize/thin_plate_spline_motion_model/tpsnn.py:511:0) (PointWiseV2)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Tactic: 0x0000000000000018 Time: 0.00337756
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Tactic: 0x0000000000000019 Time: 0.00392571
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Tactic: 0x000000000000001a Time: 0.0041704
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Tactic: 0x000000000000001b Time: 0.00440926
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Tactic: 0x000000000000001f Time: 0.00334059
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Fastest Tactic: 0x000000000000001f Time: 0.00334059
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: PWN(PWN(%kp_value_diff.1 : Tensor = aten::sub(%371, %292, %365) # /home/joy/remote/stable-diffusion-backend/personalize/thin_plate_spline_motion_model/tpsnn.py:510:0, %380 : Tensor = aten::mul(%kp_value_diff.1, %adapt_movement_scale) # /home/joy/remote/stable-diffusion-backend/personalize/thin_plate_spline_motion_model/tpsnn.py:511:0), %kp.1 : Tensor = aten::add(%380, %211, %365) # /home/joy/remote/stable-diffusion-backend/personalize/thin_plate_spline_motion_model/tpsnn.py:511:0) (PointWise)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - PointWise has no valid tactics for this config, skipping
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - >>>>>>>>>>>>>>> Chose Runner Type: PointWiseV2 Tactic: 0x000000000000001f
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Half(100,2,1), Half(100,2,1), Half(1,1,1), Half(100,2,1) -> Half(100,2,1) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: PWN(PWN(%kp_value_diff.1 : Tensor = aten::sub(%371, %292, %365) # /home/joy/remote/stable-diffusion-backend/personalize/thin_plate_spline_motion_model/tpsnn.py:510:0, %380 : Tensor = aten::mul(%kp_value_diff.1, %adapt_movement_scale) # /home/joy/remote/stable-diffusion-backend/personalize/thin_plate_spline_motion_model/tpsnn.py:511:0), %kp.1 : Tensor = aten::add(%380, %211, %365) # /home/joy/remote/stable-diffusion-backend/personalize/thin_plate_spline_motion_model/tpsnn.py:511:0) (PointWiseV2)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Tactic: 0x0000000000000000 Time: 0.00278284
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Tactic: 0x0000000000000001 Time: 0.00291209
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Tactic: 0x0000000000000002 Time: 0.00272173
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Tactic: 0x0000000000000003 Time: 0.00307494
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Tactic: 0x0000000000000004 Time: 0.0030629
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Tactic: 0x0000000000000005 Time: 0.00286523
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Tactic: 0x0000000000000006 Time: 0.0032581
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Tactic: 0x0000000000000007 Time: 0.00329778
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Tactic: 0x0000000000000008 Time: 0.00301655
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Tactic: 0x0000000000000009 Time: 0.00291525
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Tactic: 0x000000000000001c Time: 0.00274868
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Fastest Tactic: 0x0000000000000002 Time: 0.00272173
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: PWN(PWN(%kp_value_diff.1 : Tensor = aten::sub(%371, %292, %365) # /home/joy/remote/stable-diffusion-backend/personalize/thin_plate_spline_motion_model/tpsnn.py:510:0, %380 : Tensor = aten::mul(%kp_value_diff.1, %adapt_movement_scale) # /home/joy/remote/stable-diffusion-backend/personalize/thin_plate_spline_motion_model/tpsnn.py:511:0), %kp.1 : Tensor = aten::add(%380, %211, %365) # /home/joy/remote/stable-diffusion-backend/personalize/thin_plate_spline_motion_model/tpsnn.py:511:0) (PointWise)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - PointWise has no valid tactics for this config, skipping
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - >>>>>>>>>>>>>>> Chose Runner Type: PointWiseV2 Tactic: 0x0000000000000002
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Half(100:2,2,1), Half(100:2,2,1), Half(1:2,1,1), Half(100:2,2,1) -> Half(100:2,2,1) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: PWN(PWN(%kp_value_diff.1 : Tensor = aten::sub(%371, %292, %365) # /home/joy/remote/stable-diffusion-backend/personalize/thin_plate_spline_motion_model/tpsnn.py:510:0, %380 : Tensor = aten::mul(%kp_value_diff.1, %adapt_movement_scale) # /home/joy/remote/stable-diffusion-backend/personalize/thin_plate_spline_motion_model/tpsnn.py:511:0), %kp.1 : Tensor = aten::add(%380, %211, %365) # /home/joy/remote/stable-diffusion-backend/personalize/thin_plate_spline_motion_model/tpsnn.py:511:0) (PointWiseV2)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Tactic: 0x0000000000000000 Time: 0.00294671
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Tactic: 0x0000000000000001 Time: 0.00302049
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Tactic: 0x0000000000000002 Time: 0.00312085
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Tactic: 0x0000000000000003 Time: 0.00317867
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Tactic: 0x0000000000000004 Time: 0.00318802
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Tactic: 0x0000000000000005 Time: 0.00301146
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Tactic: 0x0000000000000006 Time: 0.00353976
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Tactic: 0x0000000000000007 Time: 0.00338435
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Tactic: 0x0000000000000008 Time: 0.00333349
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Tactic: 0x0000000000000009 Time: 0.0031869
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Tactic: 0x000000000000000a Time: 0.00286277
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Tactic: 0x000000000000000b Time: 0.00296411
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Tactic: 0x000000000000000c Time: 0.00289232
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Tactic: 0x000000000000000d Time: 0.00300353
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Tactic: 0x000000000000000e Time: 0.00304805
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Tactic: 0x000000000000000f Time: 0.00290375
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Tactic: 0x0000000000000010 Time: 0.00335637
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Tactic: 0x0000000000000011 Time: 0.00319675
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Tactic: 0x0000000000000012 Time: 0.00309699
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Tactic: 0x0000000000000013 Time: 0.00314288
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Tactic: 0x000000000000001c Time: 0.00297251
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Tactic: 0x000000000000001d Time: 0.00289122
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Fastest Tactic: 0x000000000000000a Time: 0.00286277
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: PWN(PWN(%kp_value_diff.1 : Tensor = aten::sub(%371, %292, %365) # /home/joy/remote/stable-diffusion-backend/personalize/thin_plate_spline_motion_model/tpsnn.py:510:0, %380 : Tensor = aten::mul(%kp_value_diff.1, %adapt_movement_scale) # /home/joy/remote/stable-diffusion-backend/personalize/thin_plate_spline_motion_model/tpsnn.py:511:0), %kp.1 : Tensor = aten::add(%380, %211, %365) # /home/joy/remote/stable-diffusion-backend/personalize/thin_plate_spline_motion_model/tpsnn.py:511:0) (PointWise)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - PointWise has no valid tactics for this config, skipping
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - >>>>>>>>>>>>>>> Chose Runner Type: PointWiseV2 Tactic: 0x000000000000000a
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Half(1:8,2,1), Half(1:8,2,1), Half(1:8,1,1), Half(1:8,2,1) -> Half(1:8,2,1) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: PWN(PWN(%kp_value_diff.1 : Tensor = aten::sub(%371, %292, %365) # /home/joy/remote/stable-diffusion-backend/personalize/thin_plate_spline_motion_model/tpsnn.py:510:0, %380 : Tensor = aten::mul(%kp_value_diff.1, %adapt_movement_scale) # /home/joy/remote/stable-diffusion-backend/personalize/thin_plate_spline_motion_model/tpsnn.py:511:0), %kp.1 : Tensor = aten::add(%380, %211, %365) # /home/joy/remote/stable-diffusion-backend/personalize/thin_plate_spline_motion_model/tpsnn.py:511:0) (PointWiseV2)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Tactic: 0x000000000000001c Time: 0.00335531
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Tactic: 0x000000000000001d Time: 0.00308662
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Tactic: 0x000000000000001e Time: 0.00287412
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Tactic: 0x000000000000001f Time: 0.00305889
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Fastest Tactic: 0x000000000000001e Time: 0.00287412
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: PWN(PWN(%kp_value_diff.1 : Tensor = aten::sub(%371, %292, %365) # /home/joy/remote/stable-diffusion-backend/personalize/thin_plate_spline_motion_model/tpsnn.py:510:0, %380 : Tensor = aten::mul(%kp_value_diff.1, %adapt_movement_scale) # /home/joy/remote/stable-diffusion-backend/personalize/thin_plate_spline_motion_model/tpsnn.py:511:0), %kp.1 : Tensor = aten::add(%380, %211, %365) # /home/joy/remote/stable-diffusion-backend/personalize/thin_plate_spline_motion_model/tpsnn.py:511:0) (PointWise)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - PointWise has no valid tactics for this config, skipping
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - >>>>>>>>>>>>>>> Chose Runner Type: PointWiseV2 Tactic: 0x000000000000001e
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Half(1:16,2,1), Half(1:16,2,1), Half(1:16,1,1), Half(1:16,2,1) -> Half(1:16,2,1) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: PWN(PWN(%kp_value_diff.1 : Tensor = aten::sub(%371, %292, %365) # /home/joy/remote/stable-diffusion-backend/personalize/thin_plate_spline_motion_model/tpsnn.py:510:0, %380 : Tensor = aten::mul(%kp_value_diff.1, %adapt_movement_scale) # /home/joy/remote/stable-diffusion-backend/personalize/thin_plate_spline_motion_model/tpsnn.py:511:0), %kp.1 : Tensor = aten::add(%380, %211, %365) # /home/joy/remote/stable-diffusion-backend/personalize/thin_plate_spline_motion_model/tpsnn.py:511:0) (PointWiseV2)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Tactic: 0x000000000000001d Time: 0.00353842
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Tactic: 0x000000000000001e Time: 0.00317162
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Tactic: 0x000000000000001f Time: 0.00299324
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Fastest Tactic: 0x000000000000001f Time: 0.00299324
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - --------------- Timing Runner: PWN(PWN(%kp_value_diff.1 : Tensor = aten::sub(%371, %292, %365) # /home/joy/remote/stable-diffusion-backend/personalize/thin_plate_spline_motion_model/tpsnn.py:510:0, %380 : Tensor = aten::mul(%kp_value_diff.1, %adapt_movement_scale) # /home/joy/remote/stable-diffusion-backend/personalize/thin_plate_spline_motion_model/tpsnn.py:511:0), %kp.1 : Tensor = aten::add(%380, %211, %365) # /home/joy/remote/stable-diffusion-backend/personalize/thin_plate_spline_motion_model/tpsnn.py:511:0) (PointWise)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - PointWise has no valid tactics for this config, skipping
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - >>>>>>>>>>>>>>> Chose Runner Type: PointWiseV2 Tactic: 0x000000000000001f
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - =============== Computing costs for 
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Float(100,2,1) -> Float(100,2,2,2,1) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Float(1,2,1) -> Float(100,2,1,2,1) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Float(1:4,2,1) -> Float(100,2,1:4,2,1) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Float(100:32,2,1) -> Float(100,2,2:32,2,1) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Half(100,2,1) -> Half(100,2,2,2,1) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Half(100:2,2,1) -> Half(100,2,2:2,2,1) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Half(1:4,2,1) -> Half(100,2,1:4,2,1) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Half(1:8,2,1) -> Half(100,2,1:8,2,1) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - *************** Autotuning format combination: Half(1:16,2,1) -> Half(100,2,1:16,2,1) ***************
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Adding reformat layer: Reformatted Input Tensor 0 to unsqueeze_node_after_{ForeignNode[(Unnamed Layer* 1) [Shuffle]...%input.115 : Tensor = aten::select(%19, %21, %16) # /home/joy/remote/stable-diffusion-backend/personalize/thin_plate_spline_motion_model/tpsnn.py:529:0]}_(Unnamed Layer* 88) [Shuffle]_output ((Unnamed Layer* 88) [Shuffle]_output) from Float(1,1) to Half(1,1)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Adding reformat layer: Reformatted Input Tensor 0 to %input.3 : Tensor = aten::_convolution(%10, %self.kp_detector.fg_encoder.conv1.weight, %5, %25, %26, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.5 : Tensor = aten::batch_norm(%input.3, %self.kp_detector.fg_encoder.bn1.weight, %self.kp_detector.fg_encoder.bn1.bias, %self.kp_detector.fg_encoder.bn1.running_mean, %self.kp_detector.fg_encoder.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %37 : Tensor = aten::relu(%input.5), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 (output_1) from Float(196608,65536,256,1) to Half(65536,1:8,256,1)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Adding reformat layer: Reformatted Input Tensor 0 to %198 : Tensor = aten::matmul(%input.113, %196) + [Freeze Tensor %199 : Tensor = trt::const(%self.kp_detector.fg_encoder.fc.bias) ] + (Unnamed Layer* 83) [Shuffle] + unsqueeze_node_after_[Freeze Tensor %199 : Tensor = trt::const(%self.kp_detector.fg_encoder.fc.bias) ] + (Unnamed Layer* 83) [Shuffle]_(Unnamed Layer* 83) [Shuffle]_output + %201 : Tensor = aten::add(%199, %198, %365) + PWN(%fg_kp.3 : Tensor = aten::sigmoid(%201), scope: __module.kp_detector # /home/joy/remote/stable-diffusion-backend/personalize/thin_plate_spline_motion_model/tpsnn.py:43:0) ((Unnamed Layer* 77) [Reduce]_output) from Half(64,1:8,64,64) to Half(512,1,1,1)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Adding reformat layer: Reformatted Input Tensor 0 to %input.231 : Tensor = aten::_convolution(%22, %self.kp_detector.fg_encoder.conv1.weight, %5, %25, %26, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.233 : Tensor = aten::batch_norm(%input.231, %self.kp_detector.fg_encoder.bn1.weight, %self.kp_detector.fg_encoder.bn1.bias, %self.kp_detector.fg_encoder.bn1.running_mean, %self.kp_detector.fg_encoder.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %295 : Tensor = aten::relu(%input.233), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 || %input.117 : Tensor = aten::_convolution(%input.115, %self.kp_detector.fg_encoder.conv1.weight, %5, %25, %26, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.119 : Tensor = aten::batch_norm(%input.117, %self.kp_detector.fg_encoder.bn1.weight, %self.kp_detector.fg_encoder.bn1.bias, %self.kp_detector.fg_encoder.bn1.running_mean, %self.kp_detector.fg_encoder.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %216 : Tensor = aten::relu(%input.119), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 ((Unnamed Layer* 10) [Shuffle]_output) from Float(196608,65536,256,1) to Half(131072,65536:2,256,1)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Adding reformat layer: Reformatted Input Tensor 0 to %fg_kp.5 : Tensor = aten::sub(%205, %208, %365), scope: __module.kp_detector # /home/joy/remote/stable-diffusion-backend/personalize/thin_plate_spline_motion_model/tpsnn.py:43:0 ((Unnamed Layer* 89) [ElementWise]_output) from Half(100,1) to Float(100,1)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Adding reformat layer: Reformatted Input Tensor 0 to %212 : Tensor = aten::reshape(%211, %213) (output_2) from Float(100,2,1) to Float(1,2,1)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Adding reformat layer: Reformatted Output Tensor 0 to %212 : Tensor = aten::reshape(%211, %213) (output_4) from Float(100,2,1,2,1) to Float(100,2,2,2,1)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Adding reformat layer: Reformatted Output Tensor 0 to %input.123 : Tensor = aten::max_pool2d(%216, %26, %25, %27, %27, %4), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.maxpool # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:780:0 ((Unnamed Layer* 99) [Pooling]_output) from Half(131072,4096:2,64,1) to Half(32768,1:8,512,8)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Adding reformat layer: Reformatted Output Tensor 0 to %input.237 : Tensor = aten::max_pool2d(%295, %26, %25, %27, %27, %4), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.maxpool # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:780:0 ((Unnamed Layer* 183) [Pooling]_output) from Half(131072,4096:2,64,1) to Half(32768,1:8,512,8)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Adding reformat layer: Reformatted Input Tensor 0 to %362 : Tensor = aten::matmul(%input.341, %361) + [Freeze Tensor %363 : Tensor = trt::const(%self.kp_detector.fg_encoder.fc.bias) ] + (Unnamed Layer* 252) [Shuffle] + unsqueeze_node_after_[Freeze Tensor %363 : Tensor = trt::const(%self.kp_detector.fg_encoder.fc.bias) ] + (Unnamed Layer* 252) [Shuffle]_(Unnamed Layer* 252) [Shuffle]_output + %364 : Tensor = aten::add(%363, %362, %365) + PWN(%fg_kp.15 : Tensor = aten::sigmoid(%364), scope: __module.kp_detector # /home/joy/remote/stable-diffusion-backend/personalize/thin_plate_spline_motion_model/tpsnn.py:43:0) ((Unnamed Layer* 246) [Reduce]_output) from Half(64,1:8,64,64) to Half(512,1,1,1)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Adding reformat layer: Reformatted Input Tensor 0 to %283 : Tensor = aten::matmul(%input.227, %282) + [Freeze Tensor %284 : Tensor = trt::const(%self.kp_detector.fg_encoder.fc.bias) ] + (Unnamed Layer* 168) [Shuffle] + unsqueeze_node_after_[Freeze Tensor %284 : Tensor = trt::const(%self.kp_detector.fg_encoder.fc.bias) ] + (Unnamed Layer* 168) [Shuffle]_(Unnamed Layer* 168) [Shuffle]_output + %285 : Tensor = aten::add(%284, %283, %365) + PWN(%fg_kp.9 : Tensor = aten::sigmoid(%285), scope: __module.kp_detector # /home/joy/remote/stable-diffusion-backend/personalize/thin_plate_spline_motion_model/tpsnn.py:43:0) ((Unnamed Layer* 162) [Reduce]_output) from Half(64,1:8,64,64) to Half(512,1,1,1)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Adding reformat layer: Reformatted Input Tensor 0 to %368 : Tensor = aten::mul(%fg_kp.15, %206), scope: __module.kp_detector # /home/joy/remote/stable-diffusion-backend/personalize/thin_plate_spline_motion_model/tpsnn.py:43:0 (%fg_kp.15 : Tensor = aten::sigmoid(%364), scope: __module.kp_detector # /home/joy/remote/stable-diffusion-backend/personalize/thin_plate_spline_motion_model/tpsnn.py:43:0_out_tensor) from Half(100,1,1,1) to Float(100,1,1,1)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Adding reformat layer: Reformatted Input Tensor 0 to %289 : Tensor = aten::mul(%fg_kp.9, %206), scope: __module.kp_detector # /home/joy/remote/stable-diffusion-backend/personalize/thin_plate_spline_motion_model/tpsnn.py:43:0 (%fg_kp.9 : Tensor = aten::sigmoid(%285), scope: __module.kp_detector # /home/joy/remote/stable-diffusion-backend/personalize/thin_plate_spline_motion_model/tpsnn.py:43:0_out_tensor) from Half(100,1,1,1) to Float(100,1,1,1)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Adding reformat layer: Reformatted Input Tensor 0 to %382 : Tensor = aten::reshape(%kp.1, %213) (output_0) from Float(100,2,1) to Float(1,2,1)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Adding reformat layer: Reformatted Output Tensor 0 to %382 : Tensor = aten::reshape(%kp.1, %213) (output_3) from Float(100,2,1,2,1) to Float(100,2,2,2,1)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - For layer unsqueeze_node_after_{ForeignNode[(Unnamed Layer* 1) [Shuffle]...%input.115 : Tensor = aten::select(%19, %21, %16) # /home/joy/remote/stable-diffusion-backend/personalize/thin_plate_spline_motion_model/tpsnn.py:529:0]}_(Unnamed Layer* 88) [Shuffle]_output a non-conforming implementation was chosen than was requested i.e. requested layer computation precision and output precision types were ignored because it resulted in faster network performance. Set BuilderFlag::kPREFER_PRECISION_CONSTRAINTS to encourage choosing a conforming implementation, or set BuilderFlag::kOBEY_PRECISION_CONSTRAINTS to require choosing a conforming implementation.
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - For layer %368 : Tensor = aten::mul(%fg_kp.15, %206), scope: __module.kp_detector # /home/joy/remote/stable-diffusion-backend/personalize/thin_plate_spline_motion_model/tpsnn.py:43:0 a non-conforming implementation was chosen than was requested i.e. requested layer computation precision and output precision types were ignored because it resulted in faster network performance. Set BuilderFlag::kPREFER_PRECISION_CONSTRAINTS to encourage choosing a conforming implementation, or set BuilderFlag::kOBEY_PRECISION_CONSTRAINTS to require choosing a conforming implementation.
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - For layer %289 : Tensor = aten::mul(%fg_kp.9, %206), scope: __module.kp_detector # /home/joy/remote/stable-diffusion-backend/personalize/thin_plate_spline_motion_model/tpsnn.py:43:0 a non-conforming implementation was chosen than was requested i.e. requested layer computation precision and output precision types were ignored because it resulted in faster network performance. Set BuilderFlag::kPREFER_PRECISION_CONSTRAINTS to encourage choosing a conforming implementation, or set BuilderFlag::kOBEY_PRECISION_CONSTRAINTS to require choosing a conforming implementation.
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Formats and tactics selection completed in 35.3149 seconds.
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - After reformat layers: 109 layers
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Total number of blocks in pre-optimized block assignment: 113
INFO: [Torch-TensorRT TorchScript Conversion Context] - Total Activation Memory: 25471377408
INFO: [Torch-TensorRT TorchScript Conversion Context] - Detected 2 inputs and 5 output network tensors.
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - %input.3 : Tensor = aten::_convolution(%10, %self.kp_detector.fg_encoder.conv1.weight, %5, %25, %26, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.5 : Tensor = aten::batch_norm(%input.3, %self.kp_detector.fg_encoder.bn1.weight, %self.kp_detector.fg_encoder.bn1.bias, %self.kp_detector.fg_encoder.bn1.running_mean, %self.kp_detector.fg_encoder.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %37 : Tensor = aten::relu(%input.5), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 Set Tactic Name: sm80_xmma_fprop_image_first_layer_f16f16_f32_f16_nhwckrsc_nhwc_hmma_k64c8r7s7_stride2x2_tile16x64x16_tensor1688 Tactic: 0x603af898ad7e47f4
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - %input.9 : Tensor = aten::max_pool2d(%37, %26, %25, %27, %27, %4), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.maxpool # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:780:0 Set Tactic Name: sm50_xmma_pooling_coalescedC_NHWC_kMAX_3_False Tactic: 0xdb415cba6b0e9137
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - %input.11 : Tensor = aten::_convolution(%input.9, %self.kp_detector.fg_encoder.layer1.0.conv1.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.0/__module.kp_detector.fg_encoder.layer1.0.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.13 : Tensor = aten::batch_norm(%input.11, %self.kp_detector.fg_encoder.layer1.0.bn1.weight, %self.kp_detector.fg_encoder.layer1.0.bn1.bias, %self.kp_detector.fg_encoder.layer1.0.bn1.running_mean, %self.kp_detector.fg_encoder.layer1.0.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.0/__module.kp_detector.fg_encoder.layer1.0.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %46 : Tensor = aten::relu(%input.13), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.0/__module.kp_detector.fg_encoder.layer1.0.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 Set Tactic Name: sm80_xmma_fprop_implicit_gemm_f16f16_f16f16_f16_nhwckrsc_nhwc_tilesize64x64x64_stage4_warpsize2x2x1_g1_tensor16x8x16_t1r3s3_aACCESS Tactic: 0x30e8a8d7a953e5e9
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - %input.17 : Tensor = aten::_convolution(%46, %self.kp_detector.fg_encoder.layer1.0.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.0/__module.kp_detector.fg_encoder.layer1.0.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.1 : Tensor = aten::batch_norm(%input.17, %self.kp_detector.fg_encoder.layer1.0.bn2.weight, %self.kp_detector.fg_encoder.layer1.0.bn2.bias, %self.kp_detector.fg_encoder.layer1.0.bn2.running_mean, %self.kp_detector.fg_encoder.layer1.0.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.0/__module.kp_detector.fg_encoder.layer1.0.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %54 : Tensor = aten::add(%out.1, %input.9, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.0 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %55 : Tensor = aten::relu(%54), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.0/__module.kp_detector.fg_encoder.layer1.0.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 Set Tactic Name: sm80_xmma_fprop_implicit_gemm_f16f16_f16f16_f16_nhwckrsc_nhwc_tilesize64x64x64_stage4_warpsize2x2x1_g1_tensor16x8x16_t1r3s3_aACCESS Tactic: 0x30e8a8d7a953e5e9
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - %input.23 : Tensor = aten::_convolution(%55, %self.kp_detector.fg_encoder.layer1.1.conv1.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.1/__module.kp_detector.fg_encoder.layer1.1.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.25 : Tensor = aten::batch_norm(%input.23, %self.kp_detector.fg_encoder.layer1.1.bn1.weight, %self.kp_detector.fg_encoder.layer1.1.bn1.bias, %self.kp_detector.fg_encoder.layer1.1.bn1.running_mean, %self.kp_detector.fg_encoder.layer1.1.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.1/__module.kp_detector.fg_encoder.layer1.1.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %63 : Tensor = aten::relu(%input.25), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.1/__module.kp_detector.fg_encoder.layer1.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 Set Tactic Name: sm80_xmma_fprop_implicit_gemm_f16f16_f16f16_f16_nhwckrsc_nhwc_tilesize64x64x64_stage4_warpsize2x2x1_g1_tensor16x8x16_t1r3s3_aACCESS Tactic: 0x30e8a8d7a953e5e9
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - %input.29 : Tensor = aten::_convolution(%63, %self.kp_detector.fg_encoder.layer1.1.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.1/__module.kp_detector.fg_encoder.layer1.1.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.3 : Tensor = aten::batch_norm(%input.29, %self.kp_detector.fg_encoder.layer1.1.bn2.weight, %self.kp_detector.fg_encoder.layer1.1.bn2.bias, %self.kp_detector.fg_encoder.layer1.1.bn2.running_mean, %self.kp_detector.fg_encoder.layer1.1.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.1/__module.kp_detector.fg_encoder.layer1.1.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %71 : Tensor = aten::add(%out.3, %55, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.1 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %72 : Tensor = aten::relu(%71), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.1/__module.kp_detector.fg_encoder.layer1.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 Set Tactic Name: sm80_xmma_fprop_implicit_gemm_f16f16_f16f16_f16_nhwckrsc_nhwc_tilesize64x64x64_stage4_warpsize2x2x1_g1_tensor16x8x16_t1r3s3_aACCESS Tactic: 0x30e8a8d7a953e5e9
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - %input.35 : Tensor = aten::_convolution(%72, %self.kp_detector.fg_encoder.layer2.0.conv1.weight, %5, %25, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.0/__module.kp_detector.fg_encoder.layer2.0.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.37 : Tensor = aten::batch_norm(%input.35, %self.kp_detector.fg_encoder.layer2.0.bn1.weight, %self.kp_detector.fg_encoder.layer2.0.bn1.bias, %self.kp_detector.fg_encoder.layer2.0.bn1.running_mean, %self.kp_detector.fg_encoder.layer2.0.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.0/__module.kp_detector.fg_encoder.layer2.0.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %80 : Tensor = aten::relu(%input.37), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.0/__module.kp_detector.fg_encoder.layer2.0.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 Set Tactic Name: sm80_xmma_fprop_implicit_gemm_indexed_f16f16_f16f16_f16_nhwckrsc_nhwc_tilesize64x32x64_stage5_warpsize2x2x1_g1_tensor16x8x16 Tactic: 0xe1ff5ad20f5c6bf6
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - %input.41 : Tensor = aten::_convolution(%80, %self.kp_detector.fg_encoder.layer2.0.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.0/__module.kp_detector.fg_encoder.layer2.0.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.5 : Tensor = aten::batch_norm(%input.41, %self.kp_detector.fg_encoder.layer2.0.bn2.weight, %self.kp_detector.fg_encoder.layer2.0.bn2.bias, %self.kp_detector.fg_encoder.layer2.0.bn2.running_mean, %self.kp_detector.fg_encoder.layer2.0.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.0/__module.kp_detector.fg_encoder.layer2.0.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 Set Tactic Name: sm80_xmma_fprop_implicit_gemm_indexed_f16f16_f16f16_f16_nhwckrsc_nhwc_tilesize64x32x64_stage5_warpsize2x2x1_g1_tensor16x8x16 Tactic: 0xe1ff5ad20f5c6bf6
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - %input.43 : Tensor = aten::_convolution(%72, %self.kp_detector.fg_encoder.layer2.0.downsample.0.weight, %5, %25, %28, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.0/__module.kp_detector.fg_encoder.layer2.0.downsample/__module.kp_detector.fg_encoder.layer2.0.downsample.0 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %identity.1 : Tensor = aten::batch_norm(%input.43, %self.kp_detector.fg_encoder.layer2.0.downsample.1.weight, %self.kp_detector.fg_encoder.layer2.0.bn2.bias, %self.kp_detector.fg_encoder.layer2.0.downsample.1.running_mean, %self.kp_detector.fg_encoder.layer2.0.downsample.1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.0/__module.kp_detector.fg_encoder.layer2.0.downsample/__module.kp_detector.fg_encoder.layer2.0.downsample.1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %94 : Tensor = aten::add(%out.5, %identity.1, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.0 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %95 : Tensor = aten::relu(%94), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.0/__module.kp_detector.fg_encoder.layer2.0.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 Set Tactic Name: sm80_xmma_fprop_implicit_gemm_f16f16_f16f16_f16_nhwckrsc_nhwc_tilesize64x32x64_stage5_warpsize2x2x1_g1_tensor16x8x16_t1r1s1 Tactic: 0x2aa016c86360697f
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - %input.49 : Tensor = aten::_convolution(%95, %self.kp_detector.fg_encoder.layer2.1.conv1.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.1/__module.kp_detector.fg_encoder.layer2.1.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.51 : Tensor = aten::batch_norm(%input.49, %self.kp_detector.fg_encoder.layer2.1.bn1.weight, %self.kp_detector.fg_encoder.layer2.1.bn1.bias, %self.kp_detector.fg_encoder.layer2.1.bn1.running_mean, %self.kp_detector.fg_encoder.layer2.1.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.1/__module.kp_detector.fg_encoder.layer2.1.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %103 : Tensor = aten::relu(%input.51), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.1/__module.kp_detector.fg_encoder.layer2.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 Set Tactic Name: sm80_xmma_fprop_implicit_gemm_indexed_f16f16_f16f16_f16_nhwckrsc_nhwc_tilesize64x32x64_stage5_warpsize2x2x1_g1_tensor16x8x16 Tactic: 0xe1ff5ad20f5c6bf6
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - %input.55 : Tensor = aten::_convolution(%103, %self.kp_detector.fg_encoder.layer2.1.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.1/__module.kp_detector.fg_encoder.layer2.1.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.7 : Tensor = aten::batch_norm(%input.55, %self.kp_detector.fg_encoder.layer2.1.bn2.weight, %self.kp_detector.fg_encoder.layer2.1.bn2.bias, %self.kp_detector.fg_encoder.layer2.1.bn2.running_mean, %self.kp_detector.fg_encoder.layer2.1.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.1/__module.kp_detector.fg_encoder.layer2.1.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %111 : Tensor = aten::add(%out.7, %95, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.1 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %112 : Tensor = aten::relu(%111), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.1/__module.kp_detector.fg_encoder.layer2.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 Set Tactic Name: sm80_xmma_fprop_implicit_gemm_indexed_f16f16_f16f16_f16_nhwckrsc_nhwc_tilesize64x32x64_stage5_warpsize2x2x1_g1_tensor16x8x16 Tactic: 0xe1ff5ad20f5c6bf6
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - %input.61 : Tensor = aten::_convolution(%112, %self.kp_detector.fg_encoder.layer3.0.conv1.weight, %5, %25, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.63 : Tensor = aten::batch_norm(%input.61, %self.kp_detector.fg_encoder.layer3.0.bn1.weight, %self.kp_detector.fg_encoder.layer3.0.bn1.bias, %self.kp_detector.fg_encoder.layer3.0.bn1.running_mean, %self.kp_detector.fg_encoder.layer3.0.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %120 : Tensor = aten::relu(%input.63), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 Set Tactic Name: sm80_xmma_fprop_implicit_gemm_f16f16_f16f16_f16_nhwckrsc_nhwc_tilesize64x32x64_stage5_warpsize2x2x1_g1_tensor16x8x16_t1r3s3 Tactic: 0xa033e20ae9f412b2
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - %input.67 : Tensor = aten::_convolution(%120, %self.kp_detector.fg_encoder.layer3.0.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.9 : Tensor = aten::batch_norm(%input.67, %self.kp_detector.fg_encoder.layer3.0.bn2.weight, %self.kp_detector.fg_encoder.layer3.0.bn2.bias, %self.kp_detector.fg_encoder.layer3.0.bn2.running_mean, %self.kp_detector.fg_encoder.layer3.0.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 Set Tactic Name: sm80_xmma_fprop_implicit_gemm_f16f16_f16f16_f16_nhwckrsc_nhwc_tilesize64x32x64_stage5_warpsize2x2x1_g1_tensor16x8x16_t1r3s3 Tactic: 0xa033e20ae9f412b2
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - %input.69 : Tensor = aten::_convolution(%112, %self.kp_detector.fg_encoder.layer3.0.downsample.0.weight, %5, %25, %28, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.downsample/__module.kp_detector.fg_encoder.layer3.0.downsample.0 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %identity.3 : Tensor = aten::batch_norm(%input.69, %self.kp_detector.fg_encoder.layer3.0.downsample.1.weight, %self.kp_detector.fg_encoder.layer3.0.bn2.bias, %self.kp_detector.fg_encoder.layer3.0.downsample.1.running_mean, %self.kp_detector.fg_encoder.layer3.0.downsample.1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.downsample/__module.kp_detector.fg_encoder.layer3.0.downsample.1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %134 : Tensor = aten::add(%out.9, %identity.3, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %135 : Tensor = aten::relu(%134), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 Set Tactic Name: sm80_xmma_fprop_implicit_gemm_f16f16_f16f16_f16_nhwckrsc_nhwc_tilesize64x32x64_stage5_warpsize2x2x1_g1_tensor16x8x16_t1r1s1 Tactic: 0x2aa016c86360697f
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - %input.75 : Tensor = aten::_convolution(%135, %self.kp_detector.fg_encoder.layer3.1.conv1.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1/__module.kp_detector.fg_encoder.layer3.1.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.77 : Tensor = aten::batch_norm(%input.75, %self.kp_detector.fg_encoder.layer3.1.bn1.weight, %self.kp_detector.fg_encoder.layer3.1.bn1.bias, %self.kp_detector.fg_encoder.layer3.1.bn1.running_mean, %self.kp_detector.fg_encoder.layer3.1.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1/__module.kp_detector.fg_encoder.layer3.1.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %143 : Tensor = aten::relu(%input.77), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1/__module.kp_detector.fg_encoder.layer3.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 Set Tactic Name: sm80_xmma_fprop_implicit_gemm_f16f16_f16f16_f16_nhwckrsc_nhwc_tilesize64x32x64_stage5_warpsize2x2x1_g1_tensor16x8x16_t1r3s3 Tactic: 0xa033e20ae9f412b2
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - %input.81 : Tensor = aten::_convolution(%143, %self.kp_detector.fg_encoder.layer3.1.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1/__module.kp_detector.fg_encoder.layer3.1.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.11 : Tensor = aten::batch_norm(%input.81, %self.kp_detector.fg_encoder.layer3.1.bn2.weight, %self.kp_detector.fg_encoder.layer3.1.bn2.bias, %self.kp_detector.fg_encoder.layer3.1.bn2.running_mean, %self.kp_detector.fg_encoder.layer3.1.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1/__module.kp_detector.fg_encoder.layer3.1.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %151 : Tensor = aten::add(%out.11, %135, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %152 : Tensor = aten::relu(%151), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1/__module.kp_detector.fg_encoder.layer3.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 Set Tactic Name: sm80_xmma_fprop_implicit_gemm_f16f16_f16f16_f16_nhwckrsc_nhwc_tilesize64x32x64_stage5_warpsize2x2x1_g1_tensor16x8x16_t1r3s3 Tactic: 0xa033e20ae9f412b2
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - %input.87 : Tensor = aten::_convolution(%152, %self.kp_detector.fg_encoder.layer4.0.conv1.weight, %5, %25, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.89 : Tensor = aten::batch_norm(%input.87, %self.kp_detector.fg_encoder.layer4.0.bn1.weight, %self.kp_detector.fg_encoder.layer4.0.bn1.bias, %self.kp_detector.fg_encoder.layer4.0.bn1.running_mean, %self.kp_detector.fg_encoder.layer4.0.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %160 : Tensor = aten::relu(%input.89), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 Set Tactic Name: sm80_xmma_fprop_implicit_gemm_f16f16_f16f16_f16_nhwckrsc_nhwc_tilesize64x32x64_stage5_warpsize2x2x1_g1_tensor16x8x16_t1r3s3 Tactic: 0xa033e20ae9f412b2
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - %input.93 : Tensor = aten::_convolution(%160, %self.kp_detector.fg_encoder.layer4.0.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.13 : Tensor = aten::batch_norm(%input.93, %self.kp_detector.fg_encoder.layer4.0.bn2.weight, %self.kp_detector.fg_encoder.layer4.0.bn2.bias, %self.kp_detector.fg_encoder.layer4.0.bn2.running_mean, %self.kp_detector.fg_encoder.layer4.0.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 Set Tactic Name: sm80_xmma_fprop_implicit_gemm_f16f16_f16f16_f16_nhwckrsc_nhwc_tilesize64x32x64_stage5_warpsize2x2x1_g1_tensor16x8x16_t1r3s3 Tactic: 0xa033e20ae9f412b2
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - %input.95 : Tensor = aten::_convolution(%152, %self.kp_detector.fg_encoder.layer4.0.downsample.0.weight, %5, %25, %28, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.downsample/__module.kp_detector.fg_encoder.layer4.0.downsample.0 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %identity.5 : Tensor = aten::batch_norm(%input.95, %self.kp_detector.fg_encoder.layer4.0.downsample.1.weight, %self.kp_detector.fg_encoder.layer4.0.bn2.bias, %self.kp_detector.fg_encoder.layer4.0.downsample.1.running_mean, %self.kp_detector.fg_encoder.layer4.0.downsample.1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.downsample/__module.kp_detector.fg_encoder.layer4.0.downsample.1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %174 : Tensor = aten::add(%out.13, %identity.5, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %175 : Tensor = aten::relu(%174), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 Set Tactic Name: sm80_xmma_fprop_implicit_gemm_f16f16_f16f16_f16_nhwckrsc_nhwc_tilesize64x32x64_stage5_warpsize2x2x1_g1_tensor16x8x16_t1r1s1 Tactic: 0x2aa016c86360697f
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - %input.101 : Tensor = aten::_convolution(%175, %self.kp_detector.fg_encoder.layer4.1.conv1.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1/__module.kp_detector.fg_encoder.layer4.1.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.103 : Tensor = aten::batch_norm(%input.101, %self.kp_detector.fg_encoder.layer4.1.bn1.weight, %self.kp_detector.fg_encoder.layer4.1.bn1.bias, %self.kp_detector.fg_encoder.layer4.1.bn1.running_mean, %self.kp_detector.fg_encoder.layer4.1.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1/__module.kp_detector.fg_encoder.layer4.1.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %183 : Tensor = aten::relu(%input.103), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1/__module.kp_detector.fg_encoder.layer4.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 Set Tactic Name: sm80_xmma_fprop_implicit_gemm_f16f16_f16f16_f16_nhwckrsc_nhwc_tilesize64x32x64_stage5_warpsize2x2x1_g1_tensor16x8x16_t1r3s3 Tactic: 0xa033e20ae9f412b2
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - %input.107 : Tensor = aten::_convolution(%183, %self.kp_detector.fg_encoder.layer4.1.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1/__module.kp_detector.fg_encoder.layer4.1.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.15 : Tensor = aten::batch_norm(%input.107, %self.kp_detector.fg_encoder.layer4.1.bn2.weight, %self.kp_detector.fg_encoder.layer4.1.bn2.bias, %self.kp_detector.fg_encoder.layer4.1.bn2.running_mean, %self.kp_detector.fg_encoder.layer4.1.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1/__module.kp_detector.fg_encoder.layer4.1.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %191 : Tensor = aten::add(%out.15, %175, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %192 : Tensor = aten::relu(%191), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1/__module.kp_detector.fg_encoder.layer4.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 Set Tactic Name: sm80_xmma_fprop_implicit_gemm_f16f16_f16f16_f16_nhwckrsc_nhwc_tilesize64x32x64_stage5_warpsize2x2x1_g1_tensor16x8x16_t1r3s3 Tactic: 0xa033e20ae9f412b2
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - %x.5 : Tensor = aten::adaptive_avg_pool2d(%192, %27), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.avgpool # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1213:0 Set Tactic Name: sm50_xmma_pooling_fw_4d_FP16FP32NHWC_Average_FastDiv_CAlign4 Tactic: 0x56d7b61f084f251e
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - %input.231 : Tensor = aten::_convolution(%22, %self.kp_detector.fg_encoder.conv1.weight, %5, %25, %26, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.233 : Tensor = aten::batch_norm(%input.231, %self.kp_detector.fg_encoder.bn1.weight, %self.kp_detector.fg_encoder.bn1.bias, %self.kp_detector.fg_encoder.bn1.running_mean, %self.kp_detector.fg_encoder.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %295 : Tensor = aten::relu(%input.233), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 || %input.117 : Tensor = aten::_convolution(%input.115, %self.kp_detector.fg_encoder.conv1.weight, %5, %25, %26, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.119 : Tensor = aten::batch_norm(%input.117, %self.kp_detector.fg_encoder.bn1.weight, %self.kp_detector.fg_encoder.bn1.bias, %self.kp_detector.fg_encoder.bn1.running_mean, %self.kp_detector.fg_encoder.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %216 : Tensor = aten::relu(%input.119), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 Set Tactic Name: ampere_fp16x2_hcudnn_fp16x2_128x64_relu_medium_nn_v1 Tactic: 0x4cfee77ea8c324db
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - %input.123 : Tensor = aten::max_pool2d(%216, %26, %25, %27, %27, %4), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.maxpool # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:780:0 Set Tactic Name: sm50_xmma_pooling_CHWPacked_NCxHW2_kMAX Tactic: 0xc8ae100b63fd8921
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - %input.125 : Tensor = aten::_convolution(%input.123, %self.kp_detector.fg_encoder.layer1.0.conv1.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.0/__module.kp_detector.fg_encoder.layer1.0.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.127 : Tensor = aten::batch_norm(%input.125, %self.kp_detector.fg_encoder.layer1.0.bn1.weight, %self.kp_detector.fg_encoder.layer1.0.bn1.bias, %self.kp_detector.fg_encoder.layer1.0.bn1.running_mean, %self.kp_detector.fg_encoder.layer1.0.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.0/__module.kp_detector.fg_encoder.layer1.0.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %220 : Tensor = aten::relu(%input.127), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.0/__module.kp_detector.fg_encoder.layer1.0.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 Set Tactic Name: sm80_xmma_fprop_implicit_gemm_f16f16_f16f16_f16_nhwckrsc_nhwc_tilesize64x64x64_stage4_warpsize2x2x1_g1_tensor16x8x16_t1r3s3_aACCESS Tactic: 0x30e8a8d7a953e5e9
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - %input.237 : Tensor = aten::max_pool2d(%295, %26, %25, %27, %27, %4), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.maxpool # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:780:0 Set Tactic Name: sm50_xmma_pooling_CHWPacked_NCxHW2_kMAX Tactic: 0xc8ae100b63fd8921
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - %input.239 : Tensor = aten::_convolution(%input.237, %self.kp_detector.fg_encoder.layer1.0.conv1.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.0/__module.kp_detector.fg_encoder.layer1.0.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.241 : Tensor = aten::batch_norm(%input.239, %self.kp_detector.fg_encoder.layer1.0.bn1.weight, %self.kp_detector.fg_encoder.layer1.0.bn1.bias, %self.kp_detector.fg_encoder.layer1.0.bn1.running_mean, %self.kp_detector.fg_encoder.layer1.0.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.0/__module.kp_detector.fg_encoder.layer1.0.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %299 : Tensor = aten::relu(%input.241), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.0/__module.kp_detector.fg_encoder.layer1.0.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 Set Tactic Name: sm80_xmma_fprop_implicit_gemm_f16f16_f16f16_f16_nhwckrsc_nhwc_tilesize64x64x64_stage4_warpsize2x2x1_g1_tensor16x8x16_t1r3s3_aACCESS Tactic: 0x30e8a8d7a953e5e9
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - %input.245 : Tensor = aten::_convolution(%299, %self.kp_detector.fg_encoder.layer1.0.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.0/__module.kp_detector.fg_encoder.layer1.0.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.33 : Tensor = aten::batch_norm(%input.245, %self.kp_detector.fg_encoder.layer1.0.bn2.weight, %self.kp_detector.fg_encoder.layer1.0.bn2.bias, %self.kp_detector.fg_encoder.layer1.0.bn2.running_mean, %self.kp_detector.fg_encoder.layer1.0.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.0/__module.kp_detector.fg_encoder.layer1.0.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %302 : Tensor = aten::add(%out.33, %input.237, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.0 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %303 : Tensor = aten::relu(%302), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.0/__module.kp_detector.fg_encoder.layer1.0.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 Set Tactic Name: sm80_xmma_fprop_implicit_gemm_f16f16_f16f16_f16_nhwckrsc_nhwc_tilesize64x64x64_stage4_warpsize2x2x1_g1_tensor16x8x16_t1r3s3_aACCESS Tactic: 0x30e8a8d7a953e5e9
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - %input.131 : Tensor = aten::_convolution(%220, %self.kp_detector.fg_encoder.layer1.0.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.0/__module.kp_detector.fg_encoder.layer1.0.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.17 : Tensor = aten::batch_norm(%input.131, %self.kp_detector.fg_encoder.layer1.0.bn2.weight, %self.kp_detector.fg_encoder.layer1.0.bn2.bias, %self.kp_detector.fg_encoder.layer1.0.bn2.running_mean, %self.kp_detector.fg_encoder.layer1.0.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.0/__module.kp_detector.fg_encoder.layer1.0.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %223 : Tensor = aten::add(%out.17, %input.123, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.0 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %224 : Tensor = aten::relu(%223), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.0/__module.kp_detector.fg_encoder.layer1.0.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 Set Tactic Name: sm80_xmma_fprop_implicit_gemm_f16f16_f16f16_f16_nhwckrsc_nhwc_tilesize64x64x64_stage4_warpsize2x2x1_g1_tensor16x8x16_t1r3s3_aACCESS Tactic: 0x30e8a8d7a953e5e9
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - %input.137 : Tensor = aten::_convolution(%224, %self.kp_detector.fg_encoder.layer1.1.conv1.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.1/__module.kp_detector.fg_encoder.layer1.1.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.139 : Tensor = aten::batch_norm(%input.137, %self.kp_detector.fg_encoder.layer1.1.bn1.weight, %self.kp_detector.fg_encoder.layer1.1.bn1.bias, %self.kp_detector.fg_encoder.layer1.1.bn1.running_mean, %self.kp_detector.fg_encoder.layer1.1.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.1/__module.kp_detector.fg_encoder.layer1.1.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %227 : Tensor = aten::relu(%input.139), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.1/__module.kp_detector.fg_encoder.layer1.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 Set Tactic Name: sm80_xmma_fprop_implicit_gemm_f16f16_f16f16_f16_nhwckrsc_nhwc_tilesize64x64x64_stage4_warpsize2x2x1_g1_tensor16x8x16_t1r3s3_aACCESS Tactic: 0x30e8a8d7a953e5e9
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - %input.251 : Tensor = aten::_convolution(%303, %self.kp_detector.fg_encoder.layer1.1.conv1.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.1/__module.kp_detector.fg_encoder.layer1.1.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.253 : Tensor = aten::batch_norm(%input.251, %self.kp_detector.fg_encoder.layer1.1.bn1.weight, %self.kp_detector.fg_encoder.layer1.1.bn1.bias, %self.kp_detector.fg_encoder.layer1.1.bn1.running_mean, %self.kp_detector.fg_encoder.layer1.1.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.1/__module.kp_detector.fg_encoder.layer1.1.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %306 : Tensor = aten::relu(%input.253), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.1/__module.kp_detector.fg_encoder.layer1.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 Set Tactic Name: sm80_xmma_fprop_implicit_gemm_f16f16_f16f16_f16_nhwckrsc_nhwc_tilesize64x64x64_stage4_warpsize2x2x1_g1_tensor16x8x16_t1r3s3_aACCESS Tactic: 0x30e8a8d7a953e5e9
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - %input.257 : Tensor = aten::_convolution(%306, %self.kp_detector.fg_encoder.layer1.1.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.1/__module.kp_detector.fg_encoder.layer1.1.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.35 : Tensor = aten::batch_norm(%input.257, %self.kp_detector.fg_encoder.layer1.1.bn2.weight, %self.kp_detector.fg_encoder.layer1.1.bn2.bias, %self.kp_detector.fg_encoder.layer1.1.bn2.running_mean, %self.kp_detector.fg_encoder.layer1.1.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.1/__module.kp_detector.fg_encoder.layer1.1.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %309 : Tensor = aten::add(%out.35, %303, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.1 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %310 : Tensor = aten::relu(%309), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.1/__module.kp_detector.fg_encoder.layer1.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 Set Tactic Name: sm80_xmma_fprop_implicit_gemm_f16f16_f16f16_f16_nhwckrsc_nhwc_tilesize64x64x64_stage4_warpsize2x2x1_g1_tensor16x8x16_t1r3s3_aACCESS Tactic: 0x30e8a8d7a953e5e9
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - %input.143 : Tensor = aten::_convolution(%227, %self.kp_detector.fg_encoder.layer1.1.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.1/__module.kp_detector.fg_encoder.layer1.1.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.19 : Tensor = aten::batch_norm(%input.143, %self.kp_detector.fg_encoder.layer1.1.bn2.weight, %self.kp_detector.fg_encoder.layer1.1.bn2.bias, %self.kp_detector.fg_encoder.layer1.1.bn2.running_mean, %self.kp_detector.fg_encoder.layer1.1.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.1/__module.kp_detector.fg_encoder.layer1.1.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %230 : Tensor = aten::add(%out.19, %224, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.1 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %231 : Tensor = aten::relu(%230), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.1/__module.kp_detector.fg_encoder.layer1.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 Set Tactic Name: sm80_xmma_fprop_implicit_gemm_f16f16_f16f16_f16_nhwckrsc_nhwc_tilesize64x64x64_stage4_warpsize2x2x1_g1_tensor16x8x16_t1r3s3_aACCESS Tactic: 0x30e8a8d7a953e5e9
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - %input.149 : Tensor = aten::_convolution(%231, %self.kp_detector.fg_encoder.layer2.0.conv1.weight, %5, %25, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.0/__module.kp_detector.fg_encoder.layer2.0.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.151 : Tensor = aten::batch_norm(%input.149, %self.kp_detector.fg_encoder.layer2.0.bn1.weight, %self.kp_detector.fg_encoder.layer2.0.bn1.bias, %self.kp_detector.fg_encoder.layer2.0.bn1.running_mean, %self.kp_detector.fg_encoder.layer2.0.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.0/__module.kp_detector.fg_encoder.layer2.0.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %234 : Tensor = aten::relu(%input.151), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.0/__module.kp_detector.fg_encoder.layer2.0.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 Set Tactic Name: sm80_xmma_fprop_implicit_gemm_indexed_f16f16_f16f16_f16_nhwckrsc_nhwc_tilesize64x32x64_stage5_warpsize2x2x1_g1_tensor16x8x16 Tactic: 0xe1ff5ad20f5c6bf6
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - %input.263 : Tensor = aten::_convolution(%310, %self.kp_detector.fg_encoder.layer2.0.conv1.weight, %5, %25, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.0/__module.kp_detector.fg_encoder.layer2.0.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.265 : Tensor = aten::batch_norm(%input.263, %self.kp_detector.fg_encoder.layer2.0.bn1.weight, %self.kp_detector.fg_encoder.layer2.0.bn1.bias, %self.kp_detector.fg_encoder.layer2.0.bn1.running_mean, %self.kp_detector.fg_encoder.layer2.0.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.0/__module.kp_detector.fg_encoder.layer2.0.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %313 : Tensor = aten::relu(%input.265), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.0/__module.kp_detector.fg_encoder.layer2.0.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 Set Tactic Name: sm80_xmma_fprop_implicit_gemm_indexed_f16f16_f16f16_f16_nhwckrsc_nhwc_tilesize64x32x64_stage5_warpsize2x2x1_g1_tensor16x8x16 Tactic: 0xe1ff5ad20f5c6bf6
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - %input.269 : Tensor = aten::_convolution(%313, %self.kp_detector.fg_encoder.layer2.0.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.0/__module.kp_detector.fg_encoder.layer2.0.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.37 : Tensor = aten::batch_norm(%input.269, %self.kp_detector.fg_encoder.layer2.0.bn2.weight, %self.kp_detector.fg_encoder.layer2.0.bn2.bias, %self.kp_detector.fg_encoder.layer2.0.bn2.running_mean, %self.kp_detector.fg_encoder.layer2.0.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.0/__module.kp_detector.fg_encoder.layer2.0.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 Set Tactic Name: sm80_xmma_fprop_implicit_gemm_indexed_f16f16_f16f16_f16_nhwckrsc_nhwc_tilesize64x32x64_stage5_warpsize2x2x1_g1_tensor16x8x16 Tactic: 0xe1ff5ad20f5c6bf6
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - %input.155 : Tensor = aten::_convolution(%234, %self.kp_detector.fg_encoder.layer2.0.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.0/__module.kp_detector.fg_encoder.layer2.0.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.21 : Tensor = aten::batch_norm(%input.155, %self.kp_detector.fg_encoder.layer2.0.bn2.weight, %self.kp_detector.fg_encoder.layer2.0.bn2.bias, %self.kp_detector.fg_encoder.layer2.0.bn2.running_mean, %self.kp_detector.fg_encoder.layer2.0.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.0/__module.kp_detector.fg_encoder.layer2.0.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 Set Tactic Name: sm80_xmma_fprop_implicit_gemm_indexed_f16f16_f16f16_f16_nhwckrsc_nhwc_tilesize64x32x64_stage5_warpsize2x2x1_g1_tensor16x8x16 Tactic: 0xe1ff5ad20f5c6bf6
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - %input.271 : Tensor = aten::_convolution(%310, %self.kp_detector.fg_encoder.layer2.0.downsample.0.weight, %5, %25, %28, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.0/__module.kp_detector.fg_encoder.layer2.0.downsample/__module.kp_detector.fg_encoder.layer2.0.downsample.0 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %identity.13 : Tensor = aten::batch_norm(%input.271, %self.kp_detector.fg_encoder.layer2.0.downsample.1.weight, %self.kp_detector.fg_encoder.layer2.0.bn2.bias, %self.kp_detector.fg_encoder.layer2.0.downsample.1.running_mean, %self.kp_detector.fg_encoder.layer2.0.downsample.1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.0/__module.kp_detector.fg_encoder.layer2.0.downsample/__module.kp_detector.fg_encoder.layer2.0.downsample.1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %318 : Tensor = aten::add(%out.37, %identity.13, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.0 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %319 : Tensor = aten::relu(%318), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.0/__module.kp_detector.fg_encoder.layer2.0.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 Set Tactic Name: sm80_xmma_fprop_implicit_gemm_f16f16_f16f16_f16_nhwckrsc_nhwc_tilesize64x32x64_stage5_warpsize2x2x1_g1_tensor16x8x16_t1r1s1 Tactic: 0x2aa016c86360697f
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - %input.157 : Tensor = aten::_convolution(%231, %self.kp_detector.fg_encoder.layer2.0.downsample.0.weight, %5, %25, %28, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.0/__module.kp_detector.fg_encoder.layer2.0.downsample/__module.kp_detector.fg_encoder.layer2.0.downsample.0 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %identity.7 : Tensor = aten::batch_norm(%input.157, %self.kp_detector.fg_encoder.layer2.0.downsample.1.weight, %self.kp_detector.fg_encoder.layer2.0.bn2.bias, %self.kp_detector.fg_encoder.layer2.0.downsample.1.running_mean, %self.kp_detector.fg_encoder.layer2.0.downsample.1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.0/__module.kp_detector.fg_encoder.layer2.0.downsample/__module.kp_detector.fg_encoder.layer2.0.downsample.1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %239 : Tensor = aten::add(%out.21, %identity.7, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.0 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %240 : Tensor = aten::relu(%239), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.0/__module.kp_detector.fg_encoder.layer2.0.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 Set Tactic Name: sm80_xmma_fprop_implicit_gemm_f16f16_f16f16_f16_nhwckrsc_nhwc_tilesize64x32x64_stage5_warpsize2x2x1_g1_tensor16x8x16_t1r1s1 Tactic: 0x2aa016c86360697f
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - %input.163 : Tensor = aten::_convolution(%240, %self.kp_detector.fg_encoder.layer2.1.conv1.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.1/__module.kp_detector.fg_encoder.layer2.1.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.165 : Tensor = aten::batch_norm(%input.163, %self.kp_detector.fg_encoder.layer2.1.bn1.weight, %self.kp_detector.fg_encoder.layer2.1.bn1.bias, %self.kp_detector.fg_encoder.layer2.1.bn1.running_mean, %self.kp_detector.fg_encoder.layer2.1.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.1/__module.kp_detector.fg_encoder.layer2.1.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %243 : Tensor = aten::relu(%input.165), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.1/__module.kp_detector.fg_encoder.layer2.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 Set Tactic Name: sm80_xmma_fprop_implicit_gemm_indexed_f16f16_f16f16_f16_nhwckrsc_nhwc_tilesize64x32x64_stage5_warpsize2x2x1_g1_tensor16x8x16 Tactic: 0xe1ff5ad20f5c6bf6
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - %input.277 : Tensor = aten::_convolution(%319, %self.kp_detector.fg_encoder.layer2.1.conv1.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.1/__module.kp_detector.fg_encoder.layer2.1.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.279 : Tensor = aten::batch_norm(%input.277, %self.kp_detector.fg_encoder.layer2.1.bn1.weight, %self.kp_detector.fg_encoder.layer2.1.bn1.bias, %self.kp_detector.fg_encoder.layer2.1.bn1.running_mean, %self.kp_detector.fg_encoder.layer2.1.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.1/__module.kp_detector.fg_encoder.layer2.1.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %322 : Tensor = aten::relu(%input.279), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.1/__module.kp_detector.fg_encoder.layer2.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 Set Tactic Name: sm80_xmma_fprop_implicit_gemm_indexed_f16f16_f16f16_f16_nhwckrsc_nhwc_tilesize64x32x64_stage5_warpsize2x2x1_g1_tensor16x8x16 Tactic: 0xe1ff5ad20f5c6bf6
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - %input.283 : Tensor = aten::_convolution(%322, %self.kp_detector.fg_encoder.layer2.1.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.1/__module.kp_detector.fg_encoder.layer2.1.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.39 : Tensor = aten::batch_norm(%input.283, %self.kp_detector.fg_encoder.layer2.1.bn2.weight, %self.kp_detector.fg_encoder.layer2.1.bn2.bias, %self.kp_detector.fg_encoder.layer2.1.bn2.running_mean, %self.kp_detector.fg_encoder.layer2.1.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.1/__module.kp_detector.fg_encoder.layer2.1.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %325 : Tensor = aten::add(%out.39, %319, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.1 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %326 : Tensor = aten::relu(%325), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.1/__module.kp_detector.fg_encoder.layer2.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 Set Tactic Name: sm80_xmma_fprop_implicit_gemm_indexed_f16f16_f16f16_f16_nhwckrsc_nhwc_tilesize64x32x64_stage5_warpsize2x2x1_g1_tensor16x8x16 Tactic: 0xe1ff5ad20f5c6bf6
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - %input.169 : Tensor = aten::_convolution(%243, %self.kp_detector.fg_encoder.layer2.1.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.1/__module.kp_detector.fg_encoder.layer2.1.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.23 : Tensor = aten::batch_norm(%input.169, %self.kp_detector.fg_encoder.layer2.1.bn2.weight, %self.kp_detector.fg_encoder.layer2.1.bn2.bias, %self.kp_detector.fg_encoder.layer2.1.bn2.running_mean, %self.kp_detector.fg_encoder.layer2.1.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.1/__module.kp_detector.fg_encoder.layer2.1.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %246 : Tensor = aten::add(%out.23, %240, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.1 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %247 : Tensor = aten::relu(%246), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.1/__module.kp_detector.fg_encoder.layer2.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 Set Tactic Name: sm80_xmma_fprop_implicit_gemm_indexed_f16f16_f16f16_f16_nhwckrsc_nhwc_tilesize64x32x64_stage5_warpsize2x2x1_g1_tensor16x8x16 Tactic: 0xe1ff5ad20f5c6bf6
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - %input.175 : Tensor = aten::_convolution(%247, %self.kp_detector.fg_encoder.layer3.0.conv1.weight, %5, %25, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.177 : Tensor = aten::batch_norm(%input.175, %self.kp_detector.fg_encoder.layer3.0.bn1.weight, %self.kp_detector.fg_encoder.layer3.0.bn1.bias, %self.kp_detector.fg_encoder.layer3.0.bn1.running_mean, %self.kp_detector.fg_encoder.layer3.0.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %250 : Tensor = aten::relu(%input.177), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 Set Tactic Name: sm80_xmma_fprop_implicit_gemm_f16f16_f16f16_f16_nhwckrsc_nhwc_tilesize64x32x64_stage5_warpsize2x2x1_g1_tensor16x8x16_t1r3s3 Tactic: 0xa033e20ae9f412b2
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - %input.289 : Tensor = aten::_convolution(%326, %self.kp_detector.fg_encoder.layer3.0.conv1.weight, %5, %25, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.291 : Tensor = aten::batch_norm(%input.289, %self.kp_detector.fg_encoder.layer3.0.bn1.weight, %self.kp_detector.fg_encoder.layer3.0.bn1.bias, %self.kp_detector.fg_encoder.layer3.0.bn1.running_mean, %self.kp_detector.fg_encoder.layer3.0.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %329 : Tensor = aten::relu(%input.291), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 Set Tactic Name: sm80_xmma_fprop_implicit_gemm_f16f16_f16f16_f16_nhwckrsc_nhwc_tilesize64x32x64_stage5_warpsize2x2x1_g1_tensor16x8x16_t1r3s3 Tactic: 0xa033e20ae9f412b2
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - %input.295 : Tensor = aten::_convolution(%329, %self.kp_detector.fg_encoder.layer3.0.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.41 : Tensor = aten::batch_norm(%input.295, %self.kp_detector.fg_encoder.layer3.0.bn2.weight, %self.kp_detector.fg_encoder.layer3.0.bn2.bias, %self.kp_detector.fg_encoder.layer3.0.bn2.running_mean, %self.kp_detector.fg_encoder.layer3.0.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 Set Tactic Name: sm80_xmma_fprop_implicit_gemm_f16f16_f16f16_f16_nhwckrsc_nhwc_tilesize64x32x64_stage5_warpsize2x2x1_g1_tensor16x8x16_t1r3s3 Tactic: 0xa033e20ae9f412b2
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - %input.181 : Tensor = aten::_convolution(%250, %self.kp_detector.fg_encoder.layer3.0.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.25 : Tensor = aten::batch_norm(%input.181, %self.kp_detector.fg_encoder.layer3.0.bn2.weight, %self.kp_detector.fg_encoder.layer3.0.bn2.bias, %self.kp_detector.fg_encoder.layer3.0.bn2.running_mean, %self.kp_detector.fg_encoder.layer3.0.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 Set Tactic Name: sm80_xmma_fprop_implicit_gemm_f16f16_f16f16_f16_nhwckrsc_nhwc_tilesize64x32x64_stage5_warpsize2x2x1_g1_tensor16x8x16_t1r3s3 Tactic: 0xa033e20ae9f412b2
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - %input.297 : Tensor = aten::_convolution(%326, %self.kp_detector.fg_encoder.layer3.0.downsample.0.weight, %5, %25, %28, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.downsample/__module.kp_detector.fg_encoder.layer3.0.downsample.0 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %identity.15 : Tensor = aten::batch_norm(%input.297, %self.kp_detector.fg_encoder.layer3.0.downsample.1.weight, %self.kp_detector.fg_encoder.layer3.0.bn2.bias, %self.kp_detector.fg_encoder.layer3.0.downsample.1.running_mean, %self.kp_detector.fg_encoder.layer3.0.downsample.1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.downsample/__module.kp_detector.fg_encoder.layer3.0.downsample.1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %334 : Tensor = aten::add(%out.41, %identity.15, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %335 : Tensor = aten::relu(%334), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 Set Tactic Name: sm80_xmma_fprop_implicit_gemm_f16f16_f16f16_f16_nhwckrsc_nhwc_tilesize64x32x64_stage5_warpsize2x2x1_g1_tensor16x8x16_t1r1s1 Tactic: 0x2aa016c86360697f
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - %input.183 : Tensor = aten::_convolution(%247, %self.kp_detector.fg_encoder.layer3.0.downsample.0.weight, %5, %25, %28, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.downsample/__module.kp_detector.fg_encoder.layer3.0.downsample.0 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %identity.9 : Tensor = aten::batch_norm(%input.183, %self.kp_detector.fg_encoder.layer3.0.downsample.1.weight, %self.kp_detector.fg_encoder.layer3.0.bn2.bias, %self.kp_detector.fg_encoder.layer3.0.downsample.1.running_mean, %self.kp_detector.fg_encoder.layer3.0.downsample.1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.downsample/__module.kp_detector.fg_encoder.layer3.0.downsample.1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %255 : Tensor = aten::add(%out.25, %identity.9, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %256 : Tensor = aten::relu(%255), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 Set Tactic Name: sm80_xmma_fprop_implicit_gemm_f16f16_f16f16_f16_nhwckrsc_nhwc_tilesize64x32x64_stage5_warpsize2x2x1_g1_tensor16x8x16_t1r1s1 Tactic: 0x2aa016c86360697f
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - %input.189 : Tensor = aten::_convolution(%256, %self.kp_detector.fg_encoder.layer3.1.conv1.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1/__module.kp_detector.fg_encoder.layer3.1.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.191 : Tensor = aten::batch_norm(%input.189, %self.kp_detector.fg_encoder.layer3.1.bn1.weight, %self.kp_detector.fg_encoder.layer3.1.bn1.bias, %self.kp_detector.fg_encoder.layer3.1.bn1.running_mean, %self.kp_detector.fg_encoder.layer3.1.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1/__module.kp_detector.fg_encoder.layer3.1.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %259 : Tensor = aten::relu(%input.191), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1/__module.kp_detector.fg_encoder.layer3.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 Set Tactic Name: sm80_xmma_fprop_implicit_gemm_f16f16_f16f16_f16_nhwckrsc_nhwc_tilesize64x32x64_stage5_warpsize2x2x1_g1_tensor16x8x16_t1r3s3 Tactic: 0xa033e20ae9f412b2
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - %input.303 : Tensor = aten::_convolution(%335, %self.kp_detector.fg_encoder.layer3.1.conv1.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1/__module.kp_detector.fg_encoder.layer3.1.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.305 : Tensor = aten::batch_norm(%input.303, %self.kp_detector.fg_encoder.layer3.1.bn1.weight, %self.kp_detector.fg_encoder.layer3.1.bn1.bias, %self.kp_detector.fg_encoder.layer3.1.bn1.running_mean, %self.kp_detector.fg_encoder.layer3.1.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1/__module.kp_detector.fg_encoder.layer3.1.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %338 : Tensor = aten::relu(%input.305), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1/__module.kp_detector.fg_encoder.layer3.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 Set Tactic Name: sm80_xmma_fprop_implicit_gemm_f16f16_f16f16_f16_nhwckrsc_nhwc_tilesize64x32x64_stage5_warpsize2x2x1_g1_tensor16x8x16_t1r3s3 Tactic: 0xa033e20ae9f412b2
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - %input.309 : Tensor = aten::_convolution(%338, %self.kp_detector.fg_encoder.layer3.1.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1/__module.kp_detector.fg_encoder.layer3.1.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.43 : Tensor = aten::batch_norm(%input.309, %self.kp_detector.fg_encoder.layer3.1.bn2.weight, %self.kp_detector.fg_encoder.layer3.1.bn2.bias, %self.kp_detector.fg_encoder.layer3.1.bn2.running_mean, %self.kp_detector.fg_encoder.layer3.1.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1/__module.kp_detector.fg_encoder.layer3.1.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %341 : Tensor = aten::add(%out.43, %335, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %342 : Tensor = aten::relu(%341), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1/__module.kp_detector.fg_encoder.layer3.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 Set Tactic Name: sm80_xmma_fprop_implicit_gemm_f16f16_f16f16_f16_nhwckrsc_nhwc_tilesize64x32x64_stage5_warpsize2x2x1_g1_tensor16x8x16_t1r3s3 Tactic: 0xa033e20ae9f412b2
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - %input.195 : Tensor = aten::_convolution(%259, %self.kp_detector.fg_encoder.layer3.1.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1/__module.kp_detector.fg_encoder.layer3.1.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.27 : Tensor = aten::batch_norm(%input.195, %self.kp_detector.fg_encoder.layer3.1.bn2.weight, %self.kp_detector.fg_encoder.layer3.1.bn2.bias, %self.kp_detector.fg_encoder.layer3.1.bn2.running_mean, %self.kp_detector.fg_encoder.layer3.1.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1/__module.kp_detector.fg_encoder.layer3.1.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %262 : Tensor = aten::add(%out.27, %256, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %263 : Tensor = aten::relu(%262), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1/__module.kp_detector.fg_encoder.layer3.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 Set Tactic Name: sm80_xmma_fprop_implicit_gemm_f16f16_f16f16_f16_nhwckrsc_nhwc_tilesize64x32x64_stage5_warpsize2x2x1_g1_tensor16x8x16_t1r3s3 Tactic: 0xa033e20ae9f412b2
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - %input.201 : Tensor = aten::_convolution(%263, %self.kp_detector.fg_encoder.layer4.0.conv1.weight, %5, %25, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.203 : Tensor = aten::batch_norm(%input.201, %self.kp_detector.fg_encoder.layer4.0.bn1.weight, %self.kp_detector.fg_encoder.layer4.0.bn1.bias, %self.kp_detector.fg_encoder.layer4.0.bn1.running_mean, %self.kp_detector.fg_encoder.layer4.0.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %266 : Tensor = aten::relu(%input.203), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 Set Tactic Name: sm80_xmma_fprop_implicit_gemm_f16f16_f16f16_f16_nhwckrsc_nhwc_tilesize64x32x64_stage5_warpsize2x2x1_g1_tensor16x8x16_t1r3s3 Tactic: 0xa033e20ae9f412b2
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - %input.315 : Tensor = aten::_convolution(%342, %self.kp_detector.fg_encoder.layer4.0.conv1.weight, %5, %25, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.317 : Tensor = aten::batch_norm(%input.315, %self.kp_detector.fg_encoder.layer4.0.bn1.weight, %self.kp_detector.fg_encoder.layer4.0.bn1.bias, %self.kp_detector.fg_encoder.layer4.0.bn1.running_mean, %self.kp_detector.fg_encoder.layer4.0.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %345 : Tensor = aten::relu(%input.317), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 Set Tactic Name: sm80_xmma_fprop_implicit_gemm_f16f16_f16f16_f16_nhwckrsc_nhwc_tilesize64x32x64_stage5_warpsize2x2x1_g1_tensor16x8x16_t1r3s3 Tactic: 0xa033e20ae9f412b2
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - %input.321 : Tensor = aten::_convolution(%345, %self.kp_detector.fg_encoder.layer4.0.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.45 : Tensor = aten::batch_norm(%input.321, %self.kp_detector.fg_encoder.layer4.0.bn2.weight, %self.kp_detector.fg_encoder.layer4.0.bn2.bias, %self.kp_detector.fg_encoder.layer4.0.bn2.running_mean, %self.kp_detector.fg_encoder.layer4.0.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 Set Tactic Name: sm80_xmma_fprop_implicit_gemm_f16f16_f16f16_f16_nhwckrsc_nhwc_tilesize64x32x64_stage5_warpsize2x2x1_g1_tensor16x8x16_t1r3s3 Tactic: 0xa033e20ae9f412b2
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - %input.207 : Tensor = aten::_convolution(%266, %self.kp_detector.fg_encoder.layer4.0.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.29 : Tensor = aten::batch_norm(%input.207, %self.kp_detector.fg_encoder.layer4.0.bn2.weight, %self.kp_detector.fg_encoder.layer4.0.bn2.bias, %self.kp_detector.fg_encoder.layer4.0.bn2.running_mean, %self.kp_detector.fg_encoder.layer4.0.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 Set Tactic Name: sm80_xmma_fprop_implicit_gemm_f16f16_f16f16_f16_nhwckrsc_nhwc_tilesize64x32x64_stage5_warpsize2x2x1_g1_tensor16x8x16_t1r3s3 Tactic: 0xa033e20ae9f412b2
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - %input.323 : Tensor = aten::_convolution(%342, %self.kp_detector.fg_encoder.layer4.0.downsample.0.weight, %5, %25, %28, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.downsample/__module.kp_detector.fg_encoder.layer4.0.downsample.0 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %identity : Tensor = aten::batch_norm(%input.323, %self.kp_detector.fg_encoder.layer4.0.downsample.1.weight, %self.kp_detector.fg_encoder.layer4.0.bn2.bias, %self.kp_detector.fg_encoder.layer4.0.downsample.1.running_mean, %self.kp_detector.fg_encoder.layer4.0.downsample.1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.downsample/__module.kp_detector.fg_encoder.layer4.0.downsample.1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %350 : Tensor = aten::add(%out.45, %identity, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %351 : Tensor = aten::relu(%350), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 Set Tactic Name: sm80_xmma_fprop_implicit_gemm_f16f16_f16f16_f16_nhwckrsc_nhwc_tilesize64x32x64_stage5_warpsize2x2x1_g1_tensor16x8x16_t1r1s1 Tactic: 0x2aa016c86360697f
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - %input.209 : Tensor = aten::_convolution(%263, %self.kp_detector.fg_encoder.layer4.0.downsample.0.weight, %5, %25, %28, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.downsample/__module.kp_detector.fg_encoder.layer4.0.downsample.0 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %identity.11 : Tensor = aten::batch_norm(%input.209, %self.kp_detector.fg_encoder.layer4.0.downsample.1.weight, %self.kp_detector.fg_encoder.layer4.0.bn2.bias, %self.kp_detector.fg_encoder.layer4.0.downsample.1.running_mean, %self.kp_detector.fg_encoder.layer4.0.downsample.1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.downsample/__module.kp_detector.fg_encoder.layer4.0.downsample.1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %271 : Tensor = aten::add(%out.29, %identity.11, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %272 : Tensor = aten::relu(%271), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 Set Tactic Name: sm80_xmma_fprop_implicit_gemm_f16f16_f16f16_f16_nhwckrsc_nhwc_tilesize64x32x64_stage5_warpsize2x2x1_g1_tensor16x8x16_t1r1s1 Tactic: 0x2aa016c86360697f
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - %input.215 : Tensor = aten::_convolution(%272, %self.kp_detector.fg_encoder.layer4.1.conv1.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1/__module.kp_detector.fg_encoder.layer4.1.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.217 : Tensor = aten::batch_norm(%input.215, %self.kp_detector.fg_encoder.layer4.1.bn1.weight, %self.kp_detector.fg_encoder.layer4.1.bn1.bias, %self.kp_detector.fg_encoder.layer4.1.bn1.running_mean, %self.kp_detector.fg_encoder.layer4.1.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1/__module.kp_detector.fg_encoder.layer4.1.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %275 : Tensor = aten::relu(%input.217), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1/__module.kp_detector.fg_encoder.layer4.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 Set Tactic Name: sm80_xmma_fprop_implicit_gemm_f16f16_f16f16_f16_nhwckrsc_nhwc_tilesize64x32x64_stage5_warpsize2x2x1_g1_tensor16x8x16_t1r3s3 Tactic: 0xa033e20ae9f412b2
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - %input.329 : Tensor = aten::_convolution(%351, %self.kp_detector.fg_encoder.layer4.1.conv1.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1/__module.kp_detector.fg_encoder.layer4.1.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.331 : Tensor = aten::batch_norm(%input.329, %self.kp_detector.fg_encoder.layer4.1.bn1.weight, %self.kp_detector.fg_encoder.layer4.1.bn1.bias, %self.kp_detector.fg_encoder.layer4.1.bn1.running_mean, %self.kp_detector.fg_encoder.layer4.1.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1/__module.kp_detector.fg_encoder.layer4.1.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %354 : Tensor = aten::relu(%input.331), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1/__module.kp_detector.fg_encoder.layer4.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 Set Tactic Name: sm80_xmma_fprop_implicit_gemm_f16f16_f16f16_f16_nhwckrsc_nhwc_tilesize64x32x64_stage5_warpsize2x2x1_g1_tensor16x8x16_t1r3s3 Tactic: 0xa033e20ae9f412b2
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - %input.335 : Tensor = aten::_convolution(%354, %self.kp_detector.fg_encoder.layer4.1.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1/__module.kp_detector.fg_encoder.layer4.1.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.47 : Tensor = aten::batch_norm(%input.335, %self.kp_detector.fg_encoder.layer4.1.bn2.weight, %self.kp_detector.fg_encoder.layer4.1.bn2.bias, %self.kp_detector.fg_encoder.layer4.1.bn2.running_mean, %self.kp_detector.fg_encoder.layer4.1.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1/__module.kp_detector.fg_encoder.layer4.1.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %357 : Tensor = aten::add(%out.47, %351, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %358 : Tensor = aten::relu(%357), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1/__module.kp_detector.fg_encoder.layer4.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 Set Tactic Name: sm80_xmma_fprop_implicit_gemm_f16f16_f16f16_f16_nhwckrsc_nhwc_tilesize64x32x64_stage5_warpsize2x2x1_g1_tensor16x8x16_t1r3s3 Tactic: 0xa033e20ae9f412b2
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - %x.9 : Tensor = aten::adaptive_avg_pool2d(%358, %27), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.avgpool # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1213:0 Set Tactic Name: sm50_xmma_pooling_fw_4d_FP16FP32NHWC_Average_FastDiv_CAlign4 Tactic: 0x56d7b61f084f251e
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - %input.221 : Tensor = aten::_convolution(%275, %self.kp_detector.fg_encoder.layer4.1.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1/__module.kp_detector.fg_encoder.layer4.1.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.31 : Tensor = aten::batch_norm(%input.221, %self.kp_detector.fg_encoder.layer4.1.bn2.weight, %self.kp_detector.fg_encoder.layer4.1.bn2.bias, %self.kp_detector.fg_encoder.layer4.1.bn2.running_mean, %self.kp_detector.fg_encoder.layer4.1.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1/__module.kp_detector.fg_encoder.layer4.1.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %278 : Tensor = aten::add(%out.31, %272, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %279 : Tensor = aten::relu(%278), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1/__module.kp_detector.fg_encoder.layer4.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 Set Tactic Name: sm80_xmma_fprop_implicit_gemm_f16f16_f16f16_f16_nhwckrsc_nhwc_tilesize64x32x64_stage5_warpsize2x2x1_g1_tensor16x8x16_t1r3s3 Tactic: 0xa033e20ae9f412b2
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - %x.7 : Tensor = aten::adaptive_avg_pool2d(%279, %27), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.avgpool # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1213:0 Set Tactic Name: sm50_xmma_pooling_fw_4d_FP16FP32NHWC_Average_FastDiv_CAlign4 Tactic: 0x56d7b61f084f251e
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Layer: {ForeignNode[(Unnamed Layer* 1) [Shuffle]...%input.115 : Tensor = aten::select(%19, %21, %16) # /home/joy/remote/stable-diffusion-backend/personalize/thin_plate_spline_motion_model/tpsnn.py:529:0]} Host Persistent: 24 Device Persistent: 0 Scratch Memory: 0
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Layer: %input.3 : Tensor = aten::_convolution(%10, %self.kp_detector.fg_encoder.conv1.weight, %5, %25, %26, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.5 : Tensor = aten::batch_norm(%input.3, %self.kp_detector.fg_encoder.bn1.weight, %self.kp_detector.fg_encoder.bn1.bias, %self.kp_detector.fg_encoder.bn1.running_mean, %self.kp_detector.fg_encoder.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %37 : Tensor = aten::relu(%input.5), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 Host Persistent: 1792 Device Persistent: 0 Scratch Memory: 0
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Layer: %input.9 : Tensor = aten::max_pool2d(%37, %26, %25, %27, %27, %4), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.maxpool # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:780:0 Host Persistent: 1280 Device Persistent: 0 Scratch Memory: 0
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Layer: %input.11 : Tensor = aten::_convolution(%input.9, %self.kp_detector.fg_encoder.layer1.0.conv1.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.0/__module.kp_detector.fg_encoder.layer1.0.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.13 : Tensor = aten::batch_norm(%input.11, %self.kp_detector.fg_encoder.layer1.0.bn1.weight, %self.kp_detector.fg_encoder.layer1.0.bn1.bias, %self.kp_detector.fg_encoder.layer1.0.bn1.running_mean, %self.kp_detector.fg_encoder.layer1.0.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.0/__module.kp_detector.fg_encoder.layer1.0.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %46 : Tensor = aten::relu(%input.13), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.0/__module.kp_detector.fg_encoder.layer1.0.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 Host Persistent: 3264 Device Persistent: 0 Scratch Memory: 0
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Layer: %input.17 : Tensor = aten::_convolution(%46, %self.kp_detector.fg_encoder.layer1.0.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.0/__module.kp_detector.fg_encoder.layer1.0.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.1 : Tensor = aten::batch_norm(%input.17, %self.kp_detector.fg_encoder.layer1.0.bn2.weight, %self.kp_detector.fg_encoder.layer1.0.bn2.bias, %self.kp_detector.fg_encoder.layer1.0.bn2.running_mean, %self.kp_detector.fg_encoder.layer1.0.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.0/__module.kp_detector.fg_encoder.layer1.0.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %54 : Tensor = aten::add(%out.1, %input.9, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.0 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %55 : Tensor = aten::relu(%54), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.0/__module.kp_detector.fg_encoder.layer1.0.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 Host Persistent: 3264 Device Persistent: 0 Scratch Memory: 0
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Layer: %input.23 : Tensor = aten::_convolution(%55, %self.kp_detector.fg_encoder.layer1.1.conv1.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.1/__module.kp_detector.fg_encoder.layer1.1.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.25 : Tensor = aten::batch_norm(%input.23, %self.kp_detector.fg_encoder.layer1.1.bn1.weight, %self.kp_detector.fg_encoder.layer1.1.bn1.bias, %self.kp_detector.fg_encoder.layer1.1.bn1.running_mean, %self.kp_detector.fg_encoder.layer1.1.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.1/__module.kp_detector.fg_encoder.layer1.1.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %63 : Tensor = aten::relu(%input.25), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.1/__module.kp_detector.fg_encoder.layer1.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 Host Persistent: 3264 Device Persistent: 0 Scratch Memory: 0
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Layer: %input.29 : Tensor = aten::_convolution(%63, %self.kp_detector.fg_encoder.layer1.1.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.1/__module.kp_detector.fg_encoder.layer1.1.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.3 : Tensor = aten::batch_norm(%input.29, %self.kp_detector.fg_encoder.layer1.1.bn2.weight, %self.kp_detector.fg_encoder.layer1.1.bn2.bias, %self.kp_detector.fg_encoder.layer1.1.bn2.running_mean, %self.kp_detector.fg_encoder.layer1.1.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.1/__module.kp_detector.fg_encoder.layer1.1.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %71 : Tensor = aten::add(%out.3, %55, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.1 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %72 : Tensor = aten::relu(%71), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.1/__module.kp_detector.fg_encoder.layer1.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 Host Persistent: 3264 Device Persistent: 0 Scratch Memory: 0
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Layer: %input.35 : Tensor = aten::_convolution(%72, %self.kp_detector.fg_encoder.layer2.0.conv1.weight, %5, %25, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.0/__module.kp_detector.fg_encoder.layer2.0.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.37 : Tensor = aten::batch_norm(%input.35, %self.kp_detector.fg_encoder.layer2.0.bn1.weight, %self.kp_detector.fg_encoder.layer2.0.bn1.bias, %self.kp_detector.fg_encoder.layer2.0.bn1.running_mean, %self.kp_detector.fg_encoder.layer2.0.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.0/__module.kp_detector.fg_encoder.layer2.0.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %80 : Tensor = aten::relu(%input.37), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.0/__module.kp_detector.fg_encoder.layer2.0.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 Host Persistent: 2624 Device Persistent: 0 Scratch Memory: 0
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Layer: %input.41 : Tensor = aten::_convolution(%80, %self.kp_detector.fg_encoder.layer2.0.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.0/__module.kp_detector.fg_encoder.layer2.0.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.5 : Tensor = aten::batch_norm(%input.41, %self.kp_detector.fg_encoder.layer2.0.bn2.weight, %self.kp_detector.fg_encoder.layer2.0.bn2.bias, %self.kp_detector.fg_encoder.layer2.0.bn2.running_mean, %self.kp_detector.fg_encoder.layer2.0.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.0/__module.kp_detector.fg_encoder.layer2.0.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 Host Persistent: 2624 Device Persistent: 0 Scratch Memory: 0
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Layer: %input.43 : Tensor = aten::_convolution(%72, %self.kp_detector.fg_encoder.layer2.0.downsample.0.weight, %5, %25, %28, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.0/__module.kp_detector.fg_encoder.layer2.0.downsample/__module.kp_detector.fg_encoder.layer2.0.downsample.0 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %identity.1 : Tensor = aten::batch_norm(%input.43, %self.kp_detector.fg_encoder.layer2.0.downsample.1.weight, %self.kp_detector.fg_encoder.layer2.0.bn2.bias, %self.kp_detector.fg_encoder.layer2.0.downsample.1.running_mean, %self.kp_detector.fg_encoder.layer2.0.downsample.1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.0/__module.kp_detector.fg_encoder.layer2.0.downsample/__module.kp_detector.fg_encoder.layer2.0.downsample.1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %94 : Tensor = aten::add(%out.5, %identity.1, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.0 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %95 : Tensor = aten::relu(%94), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.0/__module.kp_detector.fg_encoder.layer2.0.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 Host Persistent: 3264 Device Persistent: 0 Scratch Memory: 0
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Layer: %input.49 : Tensor = aten::_convolution(%95, %self.kp_detector.fg_encoder.layer2.1.conv1.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.1/__module.kp_detector.fg_encoder.layer2.1.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.51 : Tensor = aten::batch_norm(%input.49, %self.kp_detector.fg_encoder.layer2.1.bn1.weight, %self.kp_detector.fg_encoder.layer2.1.bn1.bias, %self.kp_detector.fg_encoder.layer2.1.bn1.running_mean, %self.kp_detector.fg_encoder.layer2.1.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.1/__module.kp_detector.fg_encoder.layer2.1.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %103 : Tensor = aten::relu(%input.51), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.1/__module.kp_detector.fg_encoder.layer2.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 Host Persistent: 2624 Device Persistent: 0 Scratch Memory: 0
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Layer: %input.55 : Tensor = aten::_convolution(%103, %self.kp_detector.fg_encoder.layer2.1.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.1/__module.kp_detector.fg_encoder.layer2.1.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.7 : Tensor = aten::batch_norm(%input.55, %self.kp_detector.fg_encoder.layer2.1.bn2.weight, %self.kp_detector.fg_encoder.layer2.1.bn2.bias, %self.kp_detector.fg_encoder.layer2.1.bn2.running_mean, %self.kp_detector.fg_encoder.layer2.1.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.1/__module.kp_detector.fg_encoder.layer2.1.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %111 : Tensor = aten::add(%out.7, %95, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.1 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %112 : Tensor = aten::relu(%111), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.1/__module.kp_detector.fg_encoder.layer2.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 Host Persistent: 2624 Device Persistent: 0 Scratch Memory: 0
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Layer: %input.61 : Tensor = aten::_convolution(%112, %self.kp_detector.fg_encoder.layer3.0.conv1.weight, %5, %25, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.63 : Tensor = aten::batch_norm(%input.61, %self.kp_detector.fg_encoder.layer3.0.bn1.weight, %self.kp_detector.fg_encoder.layer3.0.bn1.bias, %self.kp_detector.fg_encoder.layer3.0.bn1.running_mean, %self.kp_detector.fg_encoder.layer3.0.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %120 : Tensor = aten::relu(%input.63), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 Host Persistent: 3264 Device Persistent: 0 Scratch Memory: 0
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Layer: %input.67 : Tensor = aten::_convolution(%120, %self.kp_detector.fg_encoder.layer3.0.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.9 : Tensor = aten::batch_norm(%input.67, %self.kp_detector.fg_encoder.layer3.0.bn2.weight, %self.kp_detector.fg_encoder.layer3.0.bn2.bias, %self.kp_detector.fg_encoder.layer3.0.bn2.running_mean, %self.kp_detector.fg_encoder.layer3.0.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 Host Persistent: 3264 Device Persistent: 0 Scratch Memory: 0
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Layer: %input.69 : Tensor = aten::_convolution(%112, %self.kp_detector.fg_encoder.layer3.0.downsample.0.weight, %5, %25, %28, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.downsample/__module.kp_detector.fg_encoder.layer3.0.downsample.0 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %identity.3 : Tensor = aten::batch_norm(%input.69, %self.kp_detector.fg_encoder.layer3.0.downsample.1.weight, %self.kp_detector.fg_encoder.layer3.0.bn2.bias, %self.kp_detector.fg_encoder.layer3.0.downsample.1.running_mean, %self.kp_detector.fg_encoder.layer3.0.downsample.1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.downsample/__module.kp_detector.fg_encoder.layer3.0.downsample.1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %134 : Tensor = aten::add(%out.9, %identity.3, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %135 : Tensor = aten::relu(%134), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 Host Persistent: 3264 Device Persistent: 0 Scratch Memory: 0
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Layer: %input.75 : Tensor = aten::_convolution(%135, %self.kp_detector.fg_encoder.layer3.1.conv1.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1/__module.kp_detector.fg_encoder.layer3.1.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.77 : Tensor = aten::batch_norm(%input.75, %self.kp_detector.fg_encoder.layer3.1.bn1.weight, %self.kp_detector.fg_encoder.layer3.1.bn1.bias, %self.kp_detector.fg_encoder.layer3.1.bn1.running_mean, %self.kp_detector.fg_encoder.layer3.1.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1/__module.kp_detector.fg_encoder.layer3.1.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %143 : Tensor = aten::relu(%input.77), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1/__module.kp_detector.fg_encoder.layer3.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 Host Persistent: 3264 Device Persistent: 0 Scratch Memory: 0
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Layer: %input.81 : Tensor = aten::_convolution(%143, %self.kp_detector.fg_encoder.layer3.1.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1/__module.kp_detector.fg_encoder.layer3.1.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.11 : Tensor = aten::batch_norm(%input.81, %self.kp_detector.fg_encoder.layer3.1.bn2.weight, %self.kp_detector.fg_encoder.layer3.1.bn2.bias, %self.kp_detector.fg_encoder.layer3.1.bn2.running_mean, %self.kp_detector.fg_encoder.layer3.1.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1/__module.kp_detector.fg_encoder.layer3.1.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %151 : Tensor = aten::add(%out.11, %135, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %152 : Tensor = aten::relu(%151), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1/__module.kp_detector.fg_encoder.layer3.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 Host Persistent: 3264 Device Persistent: 0 Scratch Memory: 0
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Layer: %input.87 : Tensor = aten::_convolution(%152, %self.kp_detector.fg_encoder.layer4.0.conv1.weight, %5, %25, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.89 : Tensor = aten::batch_norm(%input.87, %self.kp_detector.fg_encoder.layer4.0.bn1.weight, %self.kp_detector.fg_encoder.layer4.0.bn1.bias, %self.kp_detector.fg_encoder.layer4.0.bn1.running_mean, %self.kp_detector.fg_encoder.layer4.0.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %160 : Tensor = aten::relu(%input.89), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 Host Persistent: 3264 Device Persistent: 0 Scratch Memory: 0
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Layer: %input.93 : Tensor = aten::_convolution(%160, %self.kp_detector.fg_encoder.layer4.0.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.13 : Tensor = aten::batch_norm(%input.93, %self.kp_detector.fg_encoder.layer4.0.bn2.weight, %self.kp_detector.fg_encoder.layer4.0.bn2.bias, %self.kp_detector.fg_encoder.layer4.0.bn2.running_mean, %self.kp_detector.fg_encoder.layer4.0.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 Host Persistent: 3264 Device Persistent: 0 Scratch Memory: 0
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Layer: %input.95 : Tensor = aten::_convolution(%152, %self.kp_detector.fg_encoder.layer4.0.downsample.0.weight, %5, %25, %28, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.downsample/__module.kp_detector.fg_encoder.layer4.0.downsample.0 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %identity.5 : Tensor = aten::batch_norm(%input.95, %self.kp_detector.fg_encoder.layer4.0.downsample.1.weight, %self.kp_detector.fg_encoder.layer4.0.bn2.bias, %self.kp_detector.fg_encoder.layer4.0.downsample.1.running_mean, %self.kp_detector.fg_encoder.layer4.0.downsample.1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.downsample/__module.kp_detector.fg_encoder.layer4.0.downsample.1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %174 : Tensor = aten::add(%out.13, %identity.5, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %175 : Tensor = aten::relu(%174), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 Host Persistent: 3264 Device Persistent: 0 Scratch Memory: 0
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Layer: %input.101 : Tensor = aten::_convolution(%175, %self.kp_detector.fg_encoder.layer4.1.conv1.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1/__module.kp_detector.fg_encoder.layer4.1.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.103 : Tensor = aten::batch_norm(%input.101, %self.kp_detector.fg_encoder.layer4.1.bn1.weight, %self.kp_detector.fg_encoder.layer4.1.bn1.bias, %self.kp_detector.fg_encoder.layer4.1.bn1.running_mean, %self.kp_detector.fg_encoder.layer4.1.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1/__module.kp_detector.fg_encoder.layer4.1.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %183 : Tensor = aten::relu(%input.103), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1/__module.kp_detector.fg_encoder.layer4.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 Host Persistent: 3264 Device Persistent: 0 Scratch Memory: 0
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Layer: %input.107 : Tensor = aten::_convolution(%183, %self.kp_detector.fg_encoder.layer4.1.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1/__module.kp_detector.fg_encoder.layer4.1.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.15 : Tensor = aten::batch_norm(%input.107, %self.kp_detector.fg_encoder.layer4.1.bn2.weight, %self.kp_detector.fg_encoder.layer4.1.bn2.bias, %self.kp_detector.fg_encoder.layer4.1.bn2.running_mean, %self.kp_detector.fg_encoder.layer4.1.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1/__module.kp_detector.fg_encoder.layer4.1.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %191 : Tensor = aten::add(%out.15, %175, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %192 : Tensor = aten::relu(%191), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1/__module.kp_detector.fg_encoder.layer4.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 Host Persistent: 3264 Device Persistent: 0 Scratch Memory: 0
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Layer: %x.5 : Tensor = aten::adaptive_avg_pool2d(%192, %27), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.avgpool # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1213:0 Host Persistent: 1408 Device Persistent: 0 Scratch Memory: 0
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Layer: %198 : Tensor = aten::matmul(%input.113, %196) + [Freeze Tensor %199 : Tensor = trt::const(%self.kp_detector.fg_encoder.fc.bias) ] + (Unnamed Layer* 83) [Shuffle] + unsqueeze_node_after_[Freeze Tensor %199 : Tensor = trt::const(%self.kp_detector.fg_encoder.fc.bias) ] + (Unnamed Layer* 83) [Shuffle]_(Unnamed Layer* 83) [Shuffle]_output + %201 : Tensor = aten::add(%199, %198, %365) + PWN(%fg_kp.3 : Tensor = aten::sigmoid(%201), scope: __module.kp_detector # /home/joy/remote/stable-diffusion-backend/personalize/thin_plate_spline_motion_model/tpsnn.py:43:0) Host Persistent: 340 Device Persistent: 0 Scratch Memory: 0
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Layer: %input.231 : Tensor = aten::_convolution(%22, %self.kp_detector.fg_encoder.conv1.weight, %5, %25, %26, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.233 : Tensor = aten::batch_norm(%input.231, %self.kp_detector.fg_encoder.bn1.weight, %self.kp_detector.fg_encoder.bn1.bias, %self.kp_detector.fg_encoder.bn1.running_mean, %self.kp_detector.fg_encoder.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %295 : Tensor = aten::relu(%input.233), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 || %input.117 : Tensor = aten::_convolution(%input.115, %self.kp_detector.fg_encoder.conv1.weight, %5, %25, %26, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.119 : Tensor = aten::batch_norm(%input.117, %self.kp_detector.fg_encoder.bn1.weight, %self.kp_detector.fg_encoder.bn1.bias, %self.kp_detector.fg_encoder.bn1.running_mean, %self.kp_detector.fg_encoder.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %216 : Tensor = aten::relu(%input.119), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 Host Persistent: 2176 Device Persistent: 98816 Scratch Memory: 0
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Layer: %input.123 : Tensor = aten::max_pool2d(%216, %26, %25, %27, %27, %4), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.maxpool # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:780:0 Host Persistent: 1280 Device Persistent: 0 Scratch Memory: 0
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Layer: %input.125 : Tensor = aten::_convolution(%input.123, %self.kp_detector.fg_encoder.layer1.0.conv1.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.0/__module.kp_detector.fg_encoder.layer1.0.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.127 : Tensor = aten::batch_norm(%input.125, %self.kp_detector.fg_encoder.layer1.0.bn1.weight, %self.kp_detector.fg_encoder.layer1.0.bn1.bias, %self.kp_detector.fg_encoder.layer1.0.bn1.running_mean, %self.kp_detector.fg_encoder.layer1.0.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.0/__module.kp_detector.fg_encoder.layer1.0.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %220 : Tensor = aten::relu(%input.127), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.0/__module.kp_detector.fg_encoder.layer1.0.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 Host Persistent: 3264 Device Persistent: 0 Scratch Memory: 0
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Layer: %input.237 : Tensor = aten::max_pool2d(%295, %26, %25, %27, %27, %4), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.maxpool # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:780:0 Host Persistent: 1280 Device Persistent: 0 Scratch Memory: 0
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Layer: %input.239 : Tensor = aten::_convolution(%input.237, %self.kp_detector.fg_encoder.layer1.0.conv1.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.0/__module.kp_detector.fg_encoder.layer1.0.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.241 : Tensor = aten::batch_norm(%input.239, %self.kp_detector.fg_encoder.layer1.0.bn1.weight, %self.kp_detector.fg_encoder.layer1.0.bn1.bias, %self.kp_detector.fg_encoder.layer1.0.bn1.running_mean, %self.kp_detector.fg_encoder.layer1.0.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.0/__module.kp_detector.fg_encoder.layer1.0.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %299 : Tensor = aten::relu(%input.241), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.0/__module.kp_detector.fg_encoder.layer1.0.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 Host Persistent: 3264 Device Persistent: 0 Scratch Memory: 0
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Layer: %input.245 : Tensor = aten::_convolution(%299, %self.kp_detector.fg_encoder.layer1.0.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.0/__module.kp_detector.fg_encoder.layer1.0.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.33 : Tensor = aten::batch_norm(%input.245, %self.kp_detector.fg_encoder.layer1.0.bn2.weight, %self.kp_detector.fg_encoder.layer1.0.bn2.bias, %self.kp_detector.fg_encoder.layer1.0.bn2.running_mean, %self.kp_detector.fg_encoder.layer1.0.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.0/__module.kp_detector.fg_encoder.layer1.0.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %302 : Tensor = aten::add(%out.33, %input.237, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.0 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %303 : Tensor = aten::relu(%302), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.0/__module.kp_detector.fg_encoder.layer1.0.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 Host Persistent: 3264 Device Persistent: 0 Scratch Memory: 0
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Layer: %input.131 : Tensor = aten::_convolution(%220, %self.kp_detector.fg_encoder.layer1.0.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.0/__module.kp_detector.fg_encoder.layer1.0.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.17 : Tensor = aten::batch_norm(%input.131, %self.kp_detector.fg_encoder.layer1.0.bn2.weight, %self.kp_detector.fg_encoder.layer1.0.bn2.bias, %self.kp_detector.fg_encoder.layer1.0.bn2.running_mean, %self.kp_detector.fg_encoder.layer1.0.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.0/__module.kp_detector.fg_encoder.layer1.0.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %223 : Tensor = aten::add(%out.17, %input.123, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.0 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %224 : Tensor = aten::relu(%223), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.0/__module.kp_detector.fg_encoder.layer1.0.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 Host Persistent: 3264 Device Persistent: 0 Scratch Memory: 0
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Layer: %input.137 : Tensor = aten::_convolution(%224, %self.kp_detector.fg_encoder.layer1.1.conv1.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.1/__module.kp_detector.fg_encoder.layer1.1.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.139 : Tensor = aten::batch_norm(%input.137, %self.kp_detector.fg_encoder.layer1.1.bn1.weight, %self.kp_detector.fg_encoder.layer1.1.bn1.bias, %self.kp_detector.fg_encoder.layer1.1.bn1.running_mean, %self.kp_detector.fg_encoder.layer1.1.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.1/__module.kp_detector.fg_encoder.layer1.1.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %227 : Tensor = aten::relu(%input.139), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.1/__module.kp_detector.fg_encoder.layer1.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 Host Persistent: 3264 Device Persistent: 0 Scratch Memory: 0
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Layer: %input.251 : Tensor = aten::_convolution(%303, %self.kp_detector.fg_encoder.layer1.1.conv1.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.1/__module.kp_detector.fg_encoder.layer1.1.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.253 : Tensor = aten::batch_norm(%input.251, %self.kp_detector.fg_encoder.layer1.1.bn1.weight, %self.kp_detector.fg_encoder.layer1.1.bn1.bias, %self.kp_detector.fg_encoder.layer1.1.bn1.running_mean, %self.kp_detector.fg_encoder.layer1.1.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.1/__module.kp_detector.fg_encoder.layer1.1.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %306 : Tensor = aten::relu(%input.253), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.1/__module.kp_detector.fg_encoder.layer1.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 Host Persistent: 3264 Device Persistent: 0 Scratch Memory: 0
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Layer: %input.257 : Tensor = aten::_convolution(%306, %self.kp_detector.fg_encoder.layer1.1.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.1/__module.kp_detector.fg_encoder.layer1.1.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.35 : Tensor = aten::batch_norm(%input.257, %self.kp_detector.fg_encoder.layer1.1.bn2.weight, %self.kp_detector.fg_encoder.layer1.1.bn2.bias, %self.kp_detector.fg_encoder.layer1.1.bn2.running_mean, %self.kp_detector.fg_encoder.layer1.1.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.1/__module.kp_detector.fg_encoder.layer1.1.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %309 : Tensor = aten::add(%out.35, %303, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.1 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %310 : Tensor = aten::relu(%309), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.1/__module.kp_detector.fg_encoder.layer1.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 Host Persistent: 3264 Device Persistent: 0 Scratch Memory: 0
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Layer: %input.143 : Tensor = aten::_convolution(%227, %self.kp_detector.fg_encoder.layer1.1.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.1/__module.kp_detector.fg_encoder.layer1.1.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.19 : Tensor = aten::batch_norm(%input.143, %self.kp_detector.fg_encoder.layer1.1.bn2.weight, %self.kp_detector.fg_encoder.layer1.1.bn2.bias, %self.kp_detector.fg_encoder.layer1.1.bn2.running_mean, %self.kp_detector.fg_encoder.layer1.1.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.1/__module.kp_detector.fg_encoder.layer1.1.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %230 : Tensor = aten::add(%out.19, %224, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.1 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %231 : Tensor = aten::relu(%230), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.1/__module.kp_detector.fg_encoder.layer1.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 Host Persistent: 3264 Device Persistent: 0 Scratch Memory: 0
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Layer: %input.149 : Tensor = aten::_convolution(%231, %self.kp_detector.fg_encoder.layer2.0.conv1.weight, %5, %25, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.0/__module.kp_detector.fg_encoder.layer2.0.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.151 : Tensor = aten::batch_norm(%input.149, %self.kp_detector.fg_encoder.layer2.0.bn1.weight, %self.kp_detector.fg_encoder.layer2.0.bn1.bias, %self.kp_detector.fg_encoder.layer2.0.bn1.running_mean, %self.kp_detector.fg_encoder.layer2.0.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.0/__module.kp_detector.fg_encoder.layer2.0.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %234 : Tensor = aten::relu(%input.151), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.0/__module.kp_detector.fg_encoder.layer2.0.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 Host Persistent: 2624 Device Persistent: 0 Scratch Memory: 0
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Layer: %input.263 : Tensor = aten::_convolution(%310, %self.kp_detector.fg_encoder.layer2.0.conv1.weight, %5, %25, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.0/__module.kp_detector.fg_encoder.layer2.0.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.265 : Tensor = aten::batch_norm(%input.263, %self.kp_detector.fg_encoder.layer2.0.bn1.weight, %self.kp_detector.fg_encoder.layer2.0.bn1.bias, %self.kp_detector.fg_encoder.layer2.0.bn1.running_mean, %self.kp_detector.fg_encoder.layer2.0.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.0/__module.kp_detector.fg_encoder.layer2.0.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %313 : Tensor = aten::relu(%input.265), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.0/__module.kp_detector.fg_encoder.layer2.0.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 Host Persistent: 2624 Device Persistent: 0 Scratch Memory: 0
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Layer: %input.269 : Tensor = aten::_convolution(%313, %self.kp_detector.fg_encoder.layer2.0.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.0/__module.kp_detector.fg_encoder.layer2.0.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.37 : Tensor = aten::batch_norm(%input.269, %self.kp_detector.fg_encoder.layer2.0.bn2.weight, %self.kp_detector.fg_encoder.layer2.0.bn2.bias, %self.kp_detector.fg_encoder.layer2.0.bn2.running_mean, %self.kp_detector.fg_encoder.layer2.0.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.0/__module.kp_detector.fg_encoder.layer2.0.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 Host Persistent: 2624 Device Persistent: 0 Scratch Memory: 0
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Layer: %input.155 : Tensor = aten::_convolution(%234, %self.kp_detector.fg_encoder.layer2.0.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.0/__module.kp_detector.fg_encoder.layer2.0.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.21 : Tensor = aten::batch_norm(%input.155, %self.kp_detector.fg_encoder.layer2.0.bn2.weight, %self.kp_detector.fg_encoder.layer2.0.bn2.bias, %self.kp_detector.fg_encoder.layer2.0.bn2.running_mean, %self.kp_detector.fg_encoder.layer2.0.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.0/__module.kp_detector.fg_encoder.layer2.0.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 Host Persistent: 2624 Device Persistent: 0 Scratch Memory: 0
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Layer: %input.271 : Tensor = aten::_convolution(%310, %self.kp_detector.fg_encoder.layer2.0.downsample.0.weight, %5, %25, %28, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.0/__module.kp_detector.fg_encoder.layer2.0.downsample/__module.kp_detector.fg_encoder.layer2.0.downsample.0 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %identity.13 : Tensor = aten::batch_norm(%input.271, %self.kp_detector.fg_encoder.layer2.0.downsample.1.weight, %self.kp_detector.fg_encoder.layer2.0.bn2.bias, %self.kp_detector.fg_encoder.layer2.0.downsample.1.running_mean, %self.kp_detector.fg_encoder.layer2.0.downsample.1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.0/__module.kp_detector.fg_encoder.layer2.0.downsample/__module.kp_detector.fg_encoder.layer2.0.downsample.1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %318 : Tensor = aten::add(%out.37, %identity.13, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.0 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %319 : Tensor = aten::relu(%318), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.0/__module.kp_detector.fg_encoder.layer2.0.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 Host Persistent: 3264 Device Persistent: 0 Scratch Memory: 0
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Layer: %input.157 : Tensor = aten::_convolution(%231, %self.kp_detector.fg_encoder.layer2.0.downsample.0.weight, %5, %25, %28, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.0/__module.kp_detector.fg_encoder.layer2.0.downsample/__module.kp_detector.fg_encoder.layer2.0.downsample.0 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %identity.7 : Tensor = aten::batch_norm(%input.157, %self.kp_detector.fg_encoder.layer2.0.downsample.1.weight, %self.kp_detector.fg_encoder.layer2.0.bn2.bias, %self.kp_detector.fg_encoder.layer2.0.downsample.1.running_mean, %self.kp_detector.fg_encoder.layer2.0.downsample.1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.0/__module.kp_detector.fg_encoder.layer2.0.downsample/__module.kp_detector.fg_encoder.layer2.0.downsample.1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %239 : Tensor = aten::add(%out.21, %identity.7, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.0 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %240 : Tensor = aten::relu(%239), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.0/__module.kp_detector.fg_encoder.layer2.0.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 Host Persistent: 3264 Device Persistent: 0 Scratch Memory: 0
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Layer: %input.163 : Tensor = aten::_convolution(%240, %self.kp_detector.fg_encoder.layer2.1.conv1.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.1/__module.kp_detector.fg_encoder.layer2.1.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.165 : Tensor = aten::batch_norm(%input.163, %self.kp_detector.fg_encoder.layer2.1.bn1.weight, %self.kp_detector.fg_encoder.layer2.1.bn1.bias, %self.kp_detector.fg_encoder.layer2.1.bn1.running_mean, %self.kp_detector.fg_encoder.layer2.1.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.1/__module.kp_detector.fg_encoder.layer2.1.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %243 : Tensor = aten::relu(%input.165), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.1/__module.kp_detector.fg_encoder.layer2.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 Host Persistent: 2624 Device Persistent: 0 Scratch Memory: 0
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Layer: %input.277 : Tensor = aten::_convolution(%319, %self.kp_detector.fg_encoder.layer2.1.conv1.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.1/__module.kp_detector.fg_encoder.layer2.1.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.279 : Tensor = aten::batch_norm(%input.277, %self.kp_detector.fg_encoder.layer2.1.bn1.weight, %self.kp_detector.fg_encoder.layer2.1.bn1.bias, %self.kp_detector.fg_encoder.layer2.1.bn1.running_mean, %self.kp_detector.fg_encoder.layer2.1.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.1/__module.kp_detector.fg_encoder.layer2.1.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %322 : Tensor = aten::relu(%input.279), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.1/__module.kp_detector.fg_encoder.layer2.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 Host Persistent: 2624 Device Persistent: 0 Scratch Memory: 0
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Layer: %input.283 : Tensor = aten::_convolution(%322, %self.kp_detector.fg_encoder.layer2.1.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.1/__module.kp_detector.fg_encoder.layer2.1.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.39 : Tensor = aten::batch_norm(%input.283, %self.kp_detector.fg_encoder.layer2.1.bn2.weight, %self.kp_detector.fg_encoder.layer2.1.bn2.bias, %self.kp_detector.fg_encoder.layer2.1.bn2.running_mean, %self.kp_detector.fg_encoder.layer2.1.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.1/__module.kp_detector.fg_encoder.layer2.1.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %325 : Tensor = aten::add(%out.39, %319, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.1 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %326 : Tensor = aten::relu(%325), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.1/__module.kp_detector.fg_encoder.layer2.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 Host Persistent: 2624 Device Persistent: 0 Scratch Memory: 0
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Layer: %input.169 : Tensor = aten::_convolution(%243, %self.kp_detector.fg_encoder.layer2.1.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.1/__module.kp_detector.fg_encoder.layer2.1.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.23 : Tensor = aten::batch_norm(%input.169, %self.kp_detector.fg_encoder.layer2.1.bn2.weight, %self.kp_detector.fg_encoder.layer2.1.bn2.bias, %self.kp_detector.fg_encoder.layer2.1.bn2.running_mean, %self.kp_detector.fg_encoder.layer2.1.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.1/__module.kp_detector.fg_encoder.layer2.1.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %246 : Tensor = aten::add(%out.23, %240, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.1 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %247 : Tensor = aten::relu(%246), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.1/__module.kp_detector.fg_encoder.layer2.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 Host Persistent: 2624 Device Persistent: 0 Scratch Memory: 0
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Layer: %input.175 : Tensor = aten::_convolution(%247, %self.kp_detector.fg_encoder.layer3.0.conv1.weight, %5, %25, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.177 : Tensor = aten::batch_norm(%input.175, %self.kp_detector.fg_encoder.layer3.0.bn1.weight, %self.kp_detector.fg_encoder.layer3.0.bn1.bias, %self.kp_detector.fg_encoder.layer3.0.bn1.running_mean, %self.kp_detector.fg_encoder.layer3.0.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %250 : Tensor = aten::relu(%input.177), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 Host Persistent: 3264 Device Persistent: 0 Scratch Memory: 0
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Layer: %input.289 : Tensor = aten::_convolution(%326, %self.kp_detector.fg_encoder.layer3.0.conv1.weight, %5, %25, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.291 : Tensor = aten::batch_norm(%input.289, %self.kp_detector.fg_encoder.layer3.0.bn1.weight, %self.kp_detector.fg_encoder.layer3.0.bn1.bias, %self.kp_detector.fg_encoder.layer3.0.bn1.running_mean, %self.kp_detector.fg_encoder.layer3.0.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %329 : Tensor = aten::relu(%input.291), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 Host Persistent: 3264 Device Persistent: 0 Scratch Memory: 0
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Layer: %input.295 : Tensor = aten::_convolution(%329, %self.kp_detector.fg_encoder.layer3.0.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.41 : Tensor = aten::batch_norm(%input.295, %self.kp_detector.fg_encoder.layer3.0.bn2.weight, %self.kp_detector.fg_encoder.layer3.0.bn2.bias, %self.kp_detector.fg_encoder.layer3.0.bn2.running_mean, %self.kp_detector.fg_encoder.layer3.0.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 Host Persistent: 3264 Device Persistent: 0 Scratch Memory: 0
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Layer: %input.181 : Tensor = aten::_convolution(%250, %self.kp_detector.fg_encoder.layer3.0.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.25 : Tensor = aten::batch_norm(%input.181, %self.kp_detector.fg_encoder.layer3.0.bn2.weight, %self.kp_detector.fg_encoder.layer3.0.bn2.bias, %self.kp_detector.fg_encoder.layer3.0.bn2.running_mean, %self.kp_detector.fg_encoder.layer3.0.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 Host Persistent: 3264 Device Persistent: 0 Scratch Memory: 0
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Layer: %input.297 : Tensor = aten::_convolution(%326, %self.kp_detector.fg_encoder.layer3.0.downsample.0.weight, %5, %25, %28, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.downsample/__module.kp_detector.fg_encoder.layer3.0.downsample.0 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %identity.15 : Tensor = aten::batch_norm(%input.297, %self.kp_detector.fg_encoder.layer3.0.downsample.1.weight, %self.kp_detector.fg_encoder.layer3.0.bn2.bias, %self.kp_detector.fg_encoder.layer3.0.downsample.1.running_mean, %self.kp_detector.fg_encoder.layer3.0.downsample.1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.downsample/__module.kp_detector.fg_encoder.layer3.0.downsample.1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %334 : Tensor = aten::add(%out.41, %identity.15, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %335 : Tensor = aten::relu(%334), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 Host Persistent: 3264 Device Persistent: 0 Scratch Memory: 0
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Layer: %input.183 : Tensor = aten::_convolution(%247, %self.kp_detector.fg_encoder.layer3.0.downsample.0.weight, %5, %25, %28, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.downsample/__module.kp_detector.fg_encoder.layer3.0.downsample.0 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %identity.9 : Tensor = aten::batch_norm(%input.183, %self.kp_detector.fg_encoder.layer3.0.downsample.1.weight, %self.kp_detector.fg_encoder.layer3.0.bn2.bias, %self.kp_detector.fg_encoder.layer3.0.downsample.1.running_mean, %self.kp_detector.fg_encoder.layer3.0.downsample.1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.downsample/__module.kp_detector.fg_encoder.layer3.0.downsample.1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %255 : Tensor = aten::add(%out.25, %identity.9, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %256 : Tensor = aten::relu(%255), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 Host Persistent: 3264 Device Persistent: 0 Scratch Memory: 0
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Layer: %input.189 : Tensor = aten::_convolution(%256, %self.kp_detector.fg_encoder.layer3.1.conv1.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1/__module.kp_detector.fg_encoder.layer3.1.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.191 : Tensor = aten::batch_norm(%input.189, %self.kp_detector.fg_encoder.layer3.1.bn1.weight, %self.kp_detector.fg_encoder.layer3.1.bn1.bias, %self.kp_detector.fg_encoder.layer3.1.bn1.running_mean, %self.kp_detector.fg_encoder.layer3.1.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1/__module.kp_detector.fg_encoder.layer3.1.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %259 : Tensor = aten::relu(%input.191), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1/__module.kp_detector.fg_encoder.layer3.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 Host Persistent: 3264 Device Persistent: 0 Scratch Memory: 0
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Layer: %input.303 : Tensor = aten::_convolution(%335, %self.kp_detector.fg_encoder.layer3.1.conv1.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1/__module.kp_detector.fg_encoder.layer3.1.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.305 : Tensor = aten::batch_norm(%input.303, %self.kp_detector.fg_encoder.layer3.1.bn1.weight, %self.kp_detector.fg_encoder.layer3.1.bn1.bias, %self.kp_detector.fg_encoder.layer3.1.bn1.running_mean, %self.kp_detector.fg_encoder.layer3.1.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1/__module.kp_detector.fg_encoder.layer3.1.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %338 : Tensor = aten::relu(%input.305), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1/__module.kp_detector.fg_encoder.layer3.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 Host Persistent: 3264 Device Persistent: 0 Scratch Memory: 0
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Layer: %input.309 : Tensor = aten::_convolution(%338, %self.kp_detector.fg_encoder.layer3.1.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1/__module.kp_detector.fg_encoder.layer3.1.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.43 : Tensor = aten::batch_norm(%input.309, %self.kp_detector.fg_encoder.layer3.1.bn2.weight, %self.kp_detector.fg_encoder.layer3.1.bn2.bias, %self.kp_detector.fg_encoder.layer3.1.bn2.running_mean, %self.kp_detector.fg_encoder.layer3.1.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1/__module.kp_detector.fg_encoder.layer3.1.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %341 : Tensor = aten::add(%out.43, %335, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %342 : Tensor = aten::relu(%341), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1/__module.kp_detector.fg_encoder.layer3.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 Host Persistent: 3264 Device Persistent: 0 Scratch Memory: 0
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Layer: %input.195 : Tensor = aten::_convolution(%259, %self.kp_detector.fg_encoder.layer3.1.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1/__module.kp_detector.fg_encoder.layer3.1.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.27 : Tensor = aten::batch_norm(%input.195, %self.kp_detector.fg_encoder.layer3.1.bn2.weight, %self.kp_detector.fg_encoder.layer3.1.bn2.bias, %self.kp_detector.fg_encoder.layer3.1.bn2.running_mean, %self.kp_detector.fg_encoder.layer3.1.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1/__module.kp_detector.fg_encoder.layer3.1.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %262 : Tensor = aten::add(%out.27, %256, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %263 : Tensor = aten::relu(%262), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1/__module.kp_detector.fg_encoder.layer3.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 Host Persistent: 3264 Device Persistent: 0 Scratch Memory: 0
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Layer: %input.201 : Tensor = aten::_convolution(%263, %self.kp_detector.fg_encoder.layer4.0.conv1.weight, %5, %25, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.203 : Tensor = aten::batch_norm(%input.201, %self.kp_detector.fg_encoder.layer4.0.bn1.weight, %self.kp_detector.fg_encoder.layer4.0.bn1.bias, %self.kp_detector.fg_encoder.layer4.0.bn1.running_mean, %self.kp_detector.fg_encoder.layer4.0.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %266 : Tensor = aten::relu(%input.203), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 Host Persistent: 3264 Device Persistent: 0 Scratch Memory: 0
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Layer: %input.315 : Tensor = aten::_convolution(%342, %self.kp_detector.fg_encoder.layer4.0.conv1.weight, %5, %25, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.317 : Tensor = aten::batch_norm(%input.315, %self.kp_detector.fg_encoder.layer4.0.bn1.weight, %self.kp_detector.fg_encoder.layer4.0.bn1.bias, %self.kp_detector.fg_encoder.layer4.0.bn1.running_mean, %self.kp_detector.fg_encoder.layer4.0.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %345 : Tensor = aten::relu(%input.317), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 Host Persistent: 3264 Device Persistent: 0 Scratch Memory: 0
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Layer: %input.321 : Tensor = aten::_convolution(%345, %self.kp_detector.fg_encoder.layer4.0.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.45 : Tensor = aten::batch_norm(%input.321, %self.kp_detector.fg_encoder.layer4.0.bn2.weight, %self.kp_detector.fg_encoder.layer4.0.bn2.bias, %self.kp_detector.fg_encoder.layer4.0.bn2.running_mean, %self.kp_detector.fg_encoder.layer4.0.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 Host Persistent: 3264 Device Persistent: 0 Scratch Memory: 0
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Layer: %input.207 : Tensor = aten::_convolution(%266, %self.kp_detector.fg_encoder.layer4.0.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.29 : Tensor = aten::batch_norm(%input.207, %self.kp_detector.fg_encoder.layer4.0.bn2.weight, %self.kp_detector.fg_encoder.layer4.0.bn2.bias, %self.kp_detector.fg_encoder.layer4.0.bn2.running_mean, %self.kp_detector.fg_encoder.layer4.0.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 Host Persistent: 3264 Device Persistent: 0 Scratch Memory: 0
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Layer: %input.323 : Tensor = aten::_convolution(%342, %self.kp_detector.fg_encoder.layer4.0.downsample.0.weight, %5, %25, %28, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.downsample/__module.kp_detector.fg_encoder.layer4.0.downsample.0 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %identity : Tensor = aten::batch_norm(%input.323, %self.kp_detector.fg_encoder.layer4.0.downsample.1.weight, %self.kp_detector.fg_encoder.layer4.0.bn2.bias, %self.kp_detector.fg_encoder.layer4.0.downsample.1.running_mean, %self.kp_detector.fg_encoder.layer4.0.downsample.1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.downsample/__module.kp_detector.fg_encoder.layer4.0.downsample.1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %350 : Tensor = aten::add(%out.45, %identity, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %351 : Tensor = aten::relu(%350), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 Host Persistent: 3264 Device Persistent: 0 Scratch Memory: 0
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Layer: %input.209 : Tensor = aten::_convolution(%263, %self.kp_detector.fg_encoder.layer4.0.downsample.0.weight, %5, %25, %28, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.downsample/__module.kp_detector.fg_encoder.layer4.0.downsample.0 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %identity.11 : Tensor = aten::batch_norm(%input.209, %self.kp_detector.fg_encoder.layer4.0.downsample.1.weight, %self.kp_detector.fg_encoder.layer4.0.bn2.bias, %self.kp_detector.fg_encoder.layer4.0.downsample.1.running_mean, %self.kp_detector.fg_encoder.layer4.0.downsample.1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.downsample/__module.kp_detector.fg_encoder.layer4.0.downsample.1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %271 : Tensor = aten::add(%out.29, %identity.11, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %272 : Tensor = aten::relu(%271), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 Host Persistent: 3264 Device Persistent: 0 Scratch Memory: 0
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Layer: %input.215 : Tensor = aten::_convolution(%272, %self.kp_detector.fg_encoder.layer4.1.conv1.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1/__module.kp_detector.fg_encoder.layer4.1.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.217 : Tensor = aten::batch_norm(%input.215, %self.kp_detector.fg_encoder.layer4.1.bn1.weight, %self.kp_detector.fg_encoder.layer4.1.bn1.bias, %self.kp_detector.fg_encoder.layer4.1.bn1.running_mean, %self.kp_detector.fg_encoder.layer4.1.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1/__module.kp_detector.fg_encoder.layer4.1.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %275 : Tensor = aten::relu(%input.217), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1/__module.kp_detector.fg_encoder.layer4.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 Host Persistent: 3264 Device Persistent: 0 Scratch Memory: 0
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Layer: %input.329 : Tensor = aten::_convolution(%351, %self.kp_detector.fg_encoder.layer4.1.conv1.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1/__module.kp_detector.fg_encoder.layer4.1.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.331 : Tensor = aten::batch_norm(%input.329, %self.kp_detector.fg_encoder.layer4.1.bn1.weight, %self.kp_detector.fg_encoder.layer4.1.bn1.bias, %self.kp_detector.fg_encoder.layer4.1.bn1.running_mean, %self.kp_detector.fg_encoder.layer4.1.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1/__module.kp_detector.fg_encoder.layer4.1.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %354 : Tensor = aten::relu(%input.331), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1/__module.kp_detector.fg_encoder.layer4.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 Host Persistent: 3264 Device Persistent: 0 Scratch Memory: 0
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Layer: %input.335 : Tensor = aten::_convolution(%354, %self.kp_detector.fg_encoder.layer4.1.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1/__module.kp_detector.fg_encoder.layer4.1.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.47 : Tensor = aten::batch_norm(%input.335, %self.kp_detector.fg_encoder.layer4.1.bn2.weight, %self.kp_detector.fg_encoder.layer4.1.bn2.bias, %self.kp_detector.fg_encoder.layer4.1.bn2.running_mean, %self.kp_detector.fg_encoder.layer4.1.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1/__module.kp_detector.fg_encoder.layer4.1.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %357 : Tensor = aten::add(%out.47, %351, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %358 : Tensor = aten::relu(%357), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1/__module.kp_detector.fg_encoder.layer4.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 Host Persistent: 3264 Device Persistent: 0 Scratch Memory: 0
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Layer: %x.9 : Tensor = aten::adaptive_avg_pool2d(%358, %27), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.avgpool # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1213:0 Host Persistent: 1408 Device Persistent: 0 Scratch Memory: 0
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Layer: %input.221 : Tensor = aten::_convolution(%275, %self.kp_detector.fg_encoder.layer4.1.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1/__module.kp_detector.fg_encoder.layer4.1.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.31 : Tensor = aten::batch_norm(%input.221, %self.kp_detector.fg_encoder.layer4.1.bn2.weight, %self.kp_detector.fg_encoder.layer4.1.bn2.bias, %self.kp_detector.fg_encoder.layer4.1.bn2.running_mean, %self.kp_detector.fg_encoder.layer4.1.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1/__module.kp_detector.fg_encoder.layer4.1.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %278 : Tensor = aten::add(%out.31, %272, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %279 : Tensor = aten::relu(%278), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1/__module.kp_detector.fg_encoder.layer4.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 Host Persistent: 3264 Device Persistent: 0 Scratch Memory: 0
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Layer: %x.7 : Tensor = aten::adaptive_avg_pool2d(%279, %27), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.avgpool # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1213:0 Host Persistent: 1408 Device Persistent: 0 Scratch Memory: 0
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Layer: %362 : Tensor = aten::matmul(%input.341, %361) + [Freeze Tensor %363 : Tensor = trt::const(%self.kp_detector.fg_encoder.fc.bias) ] + (Unnamed Layer* 252) [Shuffle] + unsqueeze_node_after_[Freeze Tensor %363 : Tensor = trt::const(%self.kp_detector.fg_encoder.fc.bias) ] + (Unnamed Layer* 252) [Shuffle]_(Unnamed Layer* 252) [Shuffle]_output + %364 : Tensor = aten::add(%363, %362, %365) + PWN(%fg_kp.15 : Tensor = aten::sigmoid(%364), scope: __module.kp_detector # /home/joy/remote/stable-diffusion-backend/personalize/thin_plate_spline_motion_model/tpsnn.py:43:0) Host Persistent: 340 Device Persistent: 0 Scratch Memory: 0
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Layer: %283 : Tensor = aten::matmul(%input.227, %282) + [Freeze Tensor %284 : Tensor = trt::const(%self.kp_detector.fg_encoder.fc.bias) ] + (Unnamed Layer* 168) [Shuffle] + unsqueeze_node_after_[Freeze Tensor %284 : Tensor = trt::const(%self.kp_detector.fg_encoder.fc.bias) ] + (Unnamed Layer* 168) [Shuffle]_(Unnamed Layer* 168) [Shuffle]_output + %285 : Tensor = aten::add(%284, %283, %365) + PWN(%fg_kp.9 : Tensor = aten::sigmoid(%285), scope: __module.kp_detector # /home/joy/remote/stable-diffusion-backend/personalize/thin_plate_spline_motion_model/tpsnn.py:43:0) Host Persistent: 340 Device Persistent: 0 Scratch Memory: 0
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Layer: PWN(%mean_diff : Tensor = aten::sub(%372, %374, %365) # /home/joy/remote/stable-diffusion-backend/personalize/thin_plate_spline_motion_model/tpsnn.py:503:0, %376 : Tensor = aten::pow(%mean_diff, %21) # /home/joy/remote/stable-diffusion-backend/personalize/thin_plate_spline_motion_model/tpsnn.py:503:0) Host Persistent: 436 Device Persistent: 0 Scratch Memory: 0
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Layer: PWN(PWN(%kp_value_diff.1 : Tensor = aten::sub(%371, %292, %365) # /home/joy/remote/stable-diffusion-backend/personalize/thin_plate_spline_motion_model/tpsnn.py:510:0, %380 : Tensor = aten::mul(%kp_value_diff.1, %adapt_movement_scale) # /home/joy/remote/stable-diffusion-backend/personalize/thin_plate_spline_motion_model/tpsnn.py:511:0), %kp.1 : Tensor = aten::add(%380, %211, %365) # /home/joy/remote/stable-diffusion-backend/personalize/thin_plate_spline_motion_model/tpsnn.py:511:0) Host Persistent: 580 Device Persistent: 0 Scratch Memory: 0
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Skipped printing memory information for 38 layers with 0 memory size i.e. Host Persistent + Device Persistent + Scratch Memory == 0.
INFO: [Torch-TensorRT TorchScript Conversion Context] - Total Host Persistent Memory: 192528
INFO: [Torch-TensorRT TorchScript Conversion Context] - Total Device Persistent Memory: 98816
INFO: [Torch-TensorRT TorchScript Conversion Context] - Total Scratch Memory: 0
INFO: [Torch-TensorRT TorchScript Conversion Context] - [MemUsageStats] Peak memory usage of TRT CPU/GPU memory allocators: CPU 201 MiB, GPU 1111 MiB
INFO: [Torch-TensorRT TorchScript Conversion Context] - [BlockAssignment] Started assigning block shifts. This will take 112 steps to complete.
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - STILL ALIVE: Started step 26 of 112
INFO: [Torch-TensorRT TorchScript Conversion Context] - [BlockAssignment] Algorithm ShiftNTopDown took 4.82091ms to assign 15 blocks to 112 nodes requiring 7344640 bytes.
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Total number of blocks in optimized block assignment: 15
INFO: [Torch-TensorRT TorchScript Conversion Context] - Total Activation Memory: 7344640
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Finalize: %input.3 : Tensor = aten::_convolution(%10, %self.kp_detector.fg_encoder.conv1.weight, %5, %25, %26, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.5 : Tensor = aten::batch_norm(%input.3, %self.kp_detector.fg_encoder.bn1.weight, %self.kp_detector.fg_encoder.bn1.bias, %self.kp_detector.fg_encoder.bn1.running_mean, %self.kp_detector.fg_encoder.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %37 : Tensor = aten::relu(%input.5), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 Set kernel index: 0
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Finalize: %input.9 : Tensor = aten::max_pool2d(%37, %26, %25, %27, %27, %4), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.maxpool # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:780:0 Set kernel index: 1
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Finalize: %input.11 : Tensor = aten::_convolution(%input.9, %self.kp_detector.fg_encoder.layer1.0.conv1.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.0/__module.kp_detector.fg_encoder.layer1.0.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.13 : Tensor = aten::batch_norm(%input.11, %self.kp_detector.fg_encoder.layer1.0.bn1.weight, %self.kp_detector.fg_encoder.layer1.0.bn1.bias, %self.kp_detector.fg_encoder.layer1.0.bn1.running_mean, %self.kp_detector.fg_encoder.layer1.0.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.0/__module.kp_detector.fg_encoder.layer1.0.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %46 : Tensor = aten::relu(%input.13), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.0/__module.kp_detector.fg_encoder.layer1.0.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 Set kernel index: 2
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Finalize: %input.17 : Tensor = aten::_convolution(%46, %self.kp_detector.fg_encoder.layer1.0.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.0/__module.kp_detector.fg_encoder.layer1.0.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.1 : Tensor = aten::batch_norm(%input.17, %self.kp_detector.fg_encoder.layer1.0.bn2.weight, %self.kp_detector.fg_encoder.layer1.0.bn2.bias, %self.kp_detector.fg_encoder.layer1.0.bn2.running_mean, %self.kp_detector.fg_encoder.layer1.0.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.0/__module.kp_detector.fg_encoder.layer1.0.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %54 : Tensor = aten::add(%out.1, %input.9, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.0 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %55 : Tensor = aten::relu(%54), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.0/__module.kp_detector.fg_encoder.layer1.0.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 Set kernel index: 2
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Finalize: %input.23 : Tensor = aten::_convolution(%55, %self.kp_detector.fg_encoder.layer1.1.conv1.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.1/__module.kp_detector.fg_encoder.layer1.1.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.25 : Tensor = aten::batch_norm(%input.23, %self.kp_detector.fg_encoder.layer1.1.bn1.weight, %self.kp_detector.fg_encoder.layer1.1.bn1.bias, %self.kp_detector.fg_encoder.layer1.1.bn1.running_mean, %self.kp_detector.fg_encoder.layer1.1.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.1/__module.kp_detector.fg_encoder.layer1.1.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %63 : Tensor = aten::relu(%input.25), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.1/__module.kp_detector.fg_encoder.layer1.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 Set kernel index: 2
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Finalize: %input.29 : Tensor = aten::_convolution(%63, %self.kp_detector.fg_encoder.layer1.1.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.1/__module.kp_detector.fg_encoder.layer1.1.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.3 : Tensor = aten::batch_norm(%input.29, %self.kp_detector.fg_encoder.layer1.1.bn2.weight, %self.kp_detector.fg_encoder.layer1.1.bn2.bias, %self.kp_detector.fg_encoder.layer1.1.bn2.running_mean, %self.kp_detector.fg_encoder.layer1.1.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.1/__module.kp_detector.fg_encoder.layer1.1.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %71 : Tensor = aten::add(%out.3, %55, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.1 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %72 : Tensor = aten::relu(%71), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.1/__module.kp_detector.fg_encoder.layer1.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 Set kernel index: 2
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Finalize: %input.35 : Tensor = aten::_convolution(%72, %self.kp_detector.fg_encoder.layer2.0.conv1.weight, %5, %25, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.0/__module.kp_detector.fg_encoder.layer2.0.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.37 : Tensor = aten::batch_norm(%input.35, %self.kp_detector.fg_encoder.layer2.0.bn1.weight, %self.kp_detector.fg_encoder.layer2.0.bn1.bias, %self.kp_detector.fg_encoder.layer2.0.bn1.running_mean, %self.kp_detector.fg_encoder.layer2.0.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.0/__module.kp_detector.fg_encoder.layer2.0.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %80 : Tensor = aten::relu(%input.37), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.0/__module.kp_detector.fg_encoder.layer2.0.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 Set kernel index: 3
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Finalize: %input.41 : Tensor = aten::_convolution(%80, %self.kp_detector.fg_encoder.layer2.0.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.0/__module.kp_detector.fg_encoder.layer2.0.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.5 : Tensor = aten::batch_norm(%input.41, %self.kp_detector.fg_encoder.layer2.0.bn2.weight, %self.kp_detector.fg_encoder.layer2.0.bn2.bias, %self.kp_detector.fg_encoder.layer2.0.bn2.running_mean, %self.kp_detector.fg_encoder.layer2.0.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.0/__module.kp_detector.fg_encoder.layer2.0.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 Set kernel index: 3
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Finalize: %input.43 : Tensor = aten::_convolution(%72, %self.kp_detector.fg_encoder.layer2.0.downsample.0.weight, %5, %25, %28, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.0/__module.kp_detector.fg_encoder.layer2.0.downsample/__module.kp_detector.fg_encoder.layer2.0.downsample.0 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %identity.1 : Tensor = aten::batch_norm(%input.43, %self.kp_detector.fg_encoder.layer2.0.downsample.1.weight, %self.kp_detector.fg_encoder.layer2.0.bn2.bias, %self.kp_detector.fg_encoder.layer2.0.downsample.1.running_mean, %self.kp_detector.fg_encoder.layer2.0.downsample.1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.0/__module.kp_detector.fg_encoder.layer2.0.downsample/__module.kp_detector.fg_encoder.layer2.0.downsample.1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %94 : Tensor = aten::add(%out.5, %identity.1, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.0 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %95 : Tensor = aten::relu(%94), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.0/__module.kp_detector.fg_encoder.layer2.0.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 Set kernel index: 4
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Finalize: %input.49 : Tensor = aten::_convolution(%95, %self.kp_detector.fg_encoder.layer2.1.conv1.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.1/__module.kp_detector.fg_encoder.layer2.1.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.51 : Tensor = aten::batch_norm(%input.49, %self.kp_detector.fg_encoder.layer2.1.bn1.weight, %self.kp_detector.fg_encoder.layer2.1.bn1.bias, %self.kp_detector.fg_encoder.layer2.1.bn1.running_mean, %self.kp_detector.fg_encoder.layer2.1.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.1/__module.kp_detector.fg_encoder.layer2.1.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %103 : Tensor = aten::relu(%input.51), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.1/__module.kp_detector.fg_encoder.layer2.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 Set kernel index: 3
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Finalize: %input.55 : Tensor = aten::_convolution(%103, %self.kp_detector.fg_encoder.layer2.1.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.1/__module.kp_detector.fg_encoder.layer2.1.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.7 : Tensor = aten::batch_norm(%input.55, %self.kp_detector.fg_encoder.layer2.1.bn2.weight, %self.kp_detector.fg_encoder.layer2.1.bn2.bias, %self.kp_detector.fg_encoder.layer2.1.bn2.running_mean, %self.kp_detector.fg_encoder.layer2.1.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.1/__module.kp_detector.fg_encoder.layer2.1.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %111 : Tensor = aten::add(%out.7, %95, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.1 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %112 : Tensor = aten::relu(%111), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.1/__module.kp_detector.fg_encoder.layer2.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 Set kernel index: 3
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Finalize: %input.61 : Tensor = aten::_convolution(%112, %self.kp_detector.fg_encoder.layer3.0.conv1.weight, %5, %25, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.63 : Tensor = aten::batch_norm(%input.61, %self.kp_detector.fg_encoder.layer3.0.bn1.weight, %self.kp_detector.fg_encoder.layer3.0.bn1.bias, %self.kp_detector.fg_encoder.layer3.0.bn1.running_mean, %self.kp_detector.fg_encoder.layer3.0.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %120 : Tensor = aten::relu(%input.63), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 Set kernel index: 5
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Finalize: %input.67 : Tensor = aten::_convolution(%120, %self.kp_detector.fg_encoder.layer3.0.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.9 : Tensor = aten::batch_norm(%input.67, %self.kp_detector.fg_encoder.layer3.0.bn2.weight, %self.kp_detector.fg_encoder.layer3.0.bn2.bias, %self.kp_detector.fg_encoder.layer3.0.bn2.running_mean, %self.kp_detector.fg_encoder.layer3.0.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 Set kernel index: 5
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Finalize: %input.69 : Tensor = aten::_convolution(%112, %self.kp_detector.fg_encoder.layer3.0.downsample.0.weight, %5, %25, %28, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.downsample/__module.kp_detector.fg_encoder.layer3.0.downsample.0 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %identity.3 : Tensor = aten::batch_norm(%input.69, %self.kp_detector.fg_encoder.layer3.0.downsample.1.weight, %self.kp_detector.fg_encoder.layer3.0.bn2.bias, %self.kp_detector.fg_encoder.layer3.0.downsample.1.running_mean, %self.kp_detector.fg_encoder.layer3.0.downsample.1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.downsample/__module.kp_detector.fg_encoder.layer3.0.downsample.1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %134 : Tensor = aten::add(%out.9, %identity.3, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %135 : Tensor = aten::relu(%134), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 Set kernel index: 4
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Finalize: %input.75 : Tensor = aten::_convolution(%135, %self.kp_detector.fg_encoder.layer3.1.conv1.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1/__module.kp_detector.fg_encoder.layer3.1.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.77 : Tensor = aten::batch_norm(%input.75, %self.kp_detector.fg_encoder.layer3.1.bn1.weight, %self.kp_detector.fg_encoder.layer3.1.bn1.bias, %self.kp_detector.fg_encoder.layer3.1.bn1.running_mean, %self.kp_detector.fg_encoder.layer3.1.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1/__module.kp_detector.fg_encoder.layer3.1.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %143 : Tensor = aten::relu(%input.77), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1/__module.kp_detector.fg_encoder.layer3.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 Set kernel index: 5
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Finalize: %input.81 : Tensor = aten::_convolution(%143, %self.kp_detector.fg_encoder.layer3.1.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1/__module.kp_detector.fg_encoder.layer3.1.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.11 : Tensor = aten::batch_norm(%input.81, %self.kp_detector.fg_encoder.layer3.1.bn2.weight, %self.kp_detector.fg_encoder.layer3.1.bn2.bias, %self.kp_detector.fg_encoder.layer3.1.bn2.running_mean, %self.kp_detector.fg_encoder.layer3.1.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1/__module.kp_detector.fg_encoder.layer3.1.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %151 : Tensor = aten::add(%out.11, %135, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %152 : Tensor = aten::relu(%151), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1/__module.kp_detector.fg_encoder.layer3.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 Set kernel index: 5
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Finalize: %input.87 : Tensor = aten::_convolution(%152, %self.kp_detector.fg_encoder.layer4.0.conv1.weight, %5, %25, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.89 : Tensor = aten::batch_norm(%input.87, %self.kp_detector.fg_encoder.layer4.0.bn1.weight, %self.kp_detector.fg_encoder.layer4.0.bn1.bias, %self.kp_detector.fg_encoder.layer4.0.bn1.running_mean, %self.kp_detector.fg_encoder.layer4.0.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %160 : Tensor = aten::relu(%input.89), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 Set kernel index: 5
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Finalize: %input.93 : Tensor = aten::_convolution(%160, %self.kp_detector.fg_encoder.layer4.0.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.13 : Tensor = aten::batch_norm(%input.93, %self.kp_detector.fg_encoder.layer4.0.bn2.weight, %self.kp_detector.fg_encoder.layer4.0.bn2.bias, %self.kp_detector.fg_encoder.layer4.0.bn2.running_mean, %self.kp_detector.fg_encoder.layer4.0.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 Set kernel index: 5
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Finalize: %input.95 : Tensor = aten::_convolution(%152, %self.kp_detector.fg_encoder.layer4.0.downsample.0.weight, %5, %25, %28, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.downsample/__module.kp_detector.fg_encoder.layer4.0.downsample.0 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %identity.5 : Tensor = aten::batch_norm(%input.95, %self.kp_detector.fg_encoder.layer4.0.downsample.1.weight, %self.kp_detector.fg_encoder.layer4.0.bn2.bias, %self.kp_detector.fg_encoder.layer4.0.downsample.1.running_mean, %self.kp_detector.fg_encoder.layer4.0.downsample.1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.downsample/__module.kp_detector.fg_encoder.layer4.0.downsample.1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %174 : Tensor = aten::add(%out.13, %identity.5, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %175 : Tensor = aten::relu(%174), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 Set kernel index: 4
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Finalize: %input.101 : Tensor = aten::_convolution(%175, %self.kp_detector.fg_encoder.layer4.1.conv1.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1/__module.kp_detector.fg_encoder.layer4.1.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.103 : Tensor = aten::batch_norm(%input.101, %self.kp_detector.fg_encoder.layer4.1.bn1.weight, %self.kp_detector.fg_encoder.layer4.1.bn1.bias, %self.kp_detector.fg_encoder.layer4.1.bn1.running_mean, %self.kp_detector.fg_encoder.layer4.1.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1/__module.kp_detector.fg_encoder.layer4.1.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %183 : Tensor = aten::relu(%input.103), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1/__module.kp_detector.fg_encoder.layer4.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 Set kernel index: 5
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Finalize: %input.107 : Tensor = aten::_convolution(%183, %self.kp_detector.fg_encoder.layer4.1.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1/__module.kp_detector.fg_encoder.layer4.1.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.15 : Tensor = aten::batch_norm(%input.107, %self.kp_detector.fg_encoder.layer4.1.bn2.weight, %self.kp_detector.fg_encoder.layer4.1.bn2.bias, %self.kp_detector.fg_encoder.layer4.1.bn2.running_mean, %self.kp_detector.fg_encoder.layer4.1.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1/__module.kp_detector.fg_encoder.layer4.1.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %191 : Tensor = aten::add(%out.15, %175, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %192 : Tensor = aten::relu(%191), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1/__module.kp_detector.fg_encoder.layer4.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 Set kernel index: 5
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Finalize: %x.5 : Tensor = aten::adaptive_avg_pool2d(%192, %27), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.avgpool # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1213:0 Set kernel index: 6
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Finalize: %input.231 : Tensor = aten::_convolution(%22, %self.kp_detector.fg_encoder.conv1.weight, %5, %25, %26, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.233 : Tensor = aten::batch_norm(%input.231, %self.kp_detector.fg_encoder.bn1.weight, %self.kp_detector.fg_encoder.bn1.bias, %self.kp_detector.fg_encoder.bn1.running_mean, %self.kp_detector.fg_encoder.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %295 : Tensor = aten::relu(%input.233), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 || %input.117 : Tensor = aten::_convolution(%input.115, %self.kp_detector.fg_encoder.conv1.weight, %5, %25, %26, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.119 : Tensor = aten::batch_norm(%input.117, %self.kp_detector.fg_encoder.bn1.weight, %self.kp_detector.fg_encoder.bn1.bias, %self.kp_detector.fg_encoder.bn1.running_mean, %self.kp_detector.fg_encoder.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %216 : Tensor = aten::relu(%input.119), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 Set kernel index: 7
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Finalize: %input.123 : Tensor = aten::max_pool2d(%216, %26, %25, %27, %27, %4), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.maxpool # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:780:0 Set kernel index: 8
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Finalize: %input.125 : Tensor = aten::_convolution(%input.123, %self.kp_detector.fg_encoder.layer1.0.conv1.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.0/__module.kp_detector.fg_encoder.layer1.0.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.127 : Tensor = aten::batch_norm(%input.125, %self.kp_detector.fg_encoder.layer1.0.bn1.weight, %self.kp_detector.fg_encoder.layer1.0.bn1.bias, %self.kp_detector.fg_encoder.layer1.0.bn1.running_mean, %self.kp_detector.fg_encoder.layer1.0.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.0/__module.kp_detector.fg_encoder.layer1.0.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %220 : Tensor = aten::relu(%input.127), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.0/__module.kp_detector.fg_encoder.layer1.0.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 Set kernel index: 2
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Finalize: %input.237 : Tensor = aten::max_pool2d(%295, %26, %25, %27, %27, %4), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.maxpool # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:780:0 Set kernel index: 8
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Finalize: %input.239 : Tensor = aten::_convolution(%input.237, %self.kp_detector.fg_encoder.layer1.0.conv1.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.0/__module.kp_detector.fg_encoder.layer1.0.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.241 : Tensor = aten::batch_norm(%input.239, %self.kp_detector.fg_encoder.layer1.0.bn1.weight, %self.kp_detector.fg_encoder.layer1.0.bn1.bias, %self.kp_detector.fg_encoder.layer1.0.bn1.running_mean, %self.kp_detector.fg_encoder.layer1.0.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.0/__module.kp_detector.fg_encoder.layer1.0.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %299 : Tensor = aten::relu(%input.241), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.0/__module.kp_detector.fg_encoder.layer1.0.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 Set kernel index: 2
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Finalize: %input.245 : Tensor = aten::_convolution(%299, %self.kp_detector.fg_encoder.layer1.0.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.0/__module.kp_detector.fg_encoder.layer1.0.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.33 : Tensor = aten::batch_norm(%input.245, %self.kp_detector.fg_encoder.layer1.0.bn2.weight, %self.kp_detector.fg_encoder.layer1.0.bn2.bias, %self.kp_detector.fg_encoder.layer1.0.bn2.running_mean, %self.kp_detector.fg_encoder.layer1.0.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.0/__module.kp_detector.fg_encoder.layer1.0.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %302 : Tensor = aten::add(%out.33, %input.237, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.0 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %303 : Tensor = aten::relu(%302), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.0/__module.kp_detector.fg_encoder.layer1.0.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 Set kernel index: 2
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Finalize: %input.131 : Tensor = aten::_convolution(%220, %self.kp_detector.fg_encoder.layer1.0.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.0/__module.kp_detector.fg_encoder.layer1.0.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.17 : Tensor = aten::batch_norm(%input.131, %self.kp_detector.fg_encoder.layer1.0.bn2.weight, %self.kp_detector.fg_encoder.layer1.0.bn2.bias, %self.kp_detector.fg_encoder.layer1.0.bn2.running_mean, %self.kp_detector.fg_encoder.layer1.0.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.0/__module.kp_detector.fg_encoder.layer1.0.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %223 : Tensor = aten::add(%out.17, %input.123, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.0 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %224 : Tensor = aten::relu(%223), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.0/__module.kp_detector.fg_encoder.layer1.0.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 Set kernel index: 2
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Finalize: %input.137 : Tensor = aten::_convolution(%224, %self.kp_detector.fg_encoder.layer1.1.conv1.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.1/__module.kp_detector.fg_encoder.layer1.1.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.139 : Tensor = aten::batch_norm(%input.137, %self.kp_detector.fg_encoder.layer1.1.bn1.weight, %self.kp_detector.fg_encoder.layer1.1.bn1.bias, %self.kp_detector.fg_encoder.layer1.1.bn1.running_mean, %self.kp_detector.fg_encoder.layer1.1.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.1/__module.kp_detector.fg_encoder.layer1.1.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %227 : Tensor = aten::relu(%input.139), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.1/__module.kp_detector.fg_encoder.layer1.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 Set kernel index: 2
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Finalize: %input.251 : Tensor = aten::_convolution(%303, %self.kp_detector.fg_encoder.layer1.1.conv1.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.1/__module.kp_detector.fg_encoder.layer1.1.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.253 : Tensor = aten::batch_norm(%input.251, %self.kp_detector.fg_encoder.layer1.1.bn1.weight, %self.kp_detector.fg_encoder.layer1.1.bn1.bias, %self.kp_detector.fg_encoder.layer1.1.bn1.running_mean, %self.kp_detector.fg_encoder.layer1.1.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.1/__module.kp_detector.fg_encoder.layer1.1.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %306 : Tensor = aten::relu(%input.253), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.1/__module.kp_detector.fg_encoder.layer1.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 Set kernel index: 2
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Finalize: %input.257 : Tensor = aten::_convolution(%306, %self.kp_detector.fg_encoder.layer1.1.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.1/__module.kp_detector.fg_encoder.layer1.1.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.35 : Tensor = aten::batch_norm(%input.257, %self.kp_detector.fg_encoder.layer1.1.bn2.weight, %self.kp_detector.fg_encoder.layer1.1.bn2.bias, %self.kp_detector.fg_encoder.layer1.1.bn2.running_mean, %self.kp_detector.fg_encoder.layer1.1.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.1/__module.kp_detector.fg_encoder.layer1.1.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %309 : Tensor = aten::add(%out.35, %303, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.1 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %310 : Tensor = aten::relu(%309), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.1/__module.kp_detector.fg_encoder.layer1.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 Set kernel index: 2
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Finalize: %input.143 : Tensor = aten::_convolution(%227, %self.kp_detector.fg_encoder.layer1.1.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.1/__module.kp_detector.fg_encoder.layer1.1.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.19 : Tensor = aten::batch_norm(%input.143, %self.kp_detector.fg_encoder.layer1.1.bn2.weight, %self.kp_detector.fg_encoder.layer1.1.bn2.bias, %self.kp_detector.fg_encoder.layer1.1.bn2.running_mean, %self.kp_detector.fg_encoder.layer1.1.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.1/__module.kp_detector.fg_encoder.layer1.1.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %230 : Tensor = aten::add(%out.19, %224, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.1 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %231 : Tensor = aten::relu(%230), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.1/__module.kp_detector.fg_encoder.layer1.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 Set kernel index: 2
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Finalize: %input.149 : Tensor = aten::_convolution(%231, %self.kp_detector.fg_encoder.layer2.0.conv1.weight, %5, %25, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.0/__module.kp_detector.fg_encoder.layer2.0.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.151 : Tensor = aten::batch_norm(%input.149, %self.kp_detector.fg_encoder.layer2.0.bn1.weight, %self.kp_detector.fg_encoder.layer2.0.bn1.bias, %self.kp_detector.fg_encoder.layer2.0.bn1.running_mean, %self.kp_detector.fg_encoder.layer2.0.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.0/__module.kp_detector.fg_encoder.layer2.0.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %234 : Tensor = aten::relu(%input.151), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.0/__module.kp_detector.fg_encoder.layer2.0.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 Set kernel index: 3
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Finalize: %input.263 : Tensor = aten::_convolution(%310, %self.kp_detector.fg_encoder.layer2.0.conv1.weight, %5, %25, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.0/__module.kp_detector.fg_encoder.layer2.0.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.265 : Tensor = aten::batch_norm(%input.263, %self.kp_detector.fg_encoder.layer2.0.bn1.weight, %self.kp_detector.fg_encoder.layer2.0.bn1.bias, %self.kp_detector.fg_encoder.layer2.0.bn1.running_mean, %self.kp_detector.fg_encoder.layer2.0.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.0/__module.kp_detector.fg_encoder.layer2.0.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %313 : Tensor = aten::relu(%input.265), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.0/__module.kp_detector.fg_encoder.layer2.0.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 Set kernel index: 3
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Finalize: %input.269 : Tensor = aten::_convolution(%313, %self.kp_detector.fg_encoder.layer2.0.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.0/__module.kp_detector.fg_encoder.layer2.0.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.37 : Tensor = aten::batch_norm(%input.269, %self.kp_detector.fg_encoder.layer2.0.bn2.weight, %self.kp_detector.fg_encoder.layer2.0.bn2.bias, %self.kp_detector.fg_encoder.layer2.0.bn2.running_mean, %self.kp_detector.fg_encoder.layer2.0.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.0/__module.kp_detector.fg_encoder.layer2.0.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 Set kernel index: 3
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Finalize: %input.155 : Tensor = aten::_convolution(%234, %self.kp_detector.fg_encoder.layer2.0.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.0/__module.kp_detector.fg_encoder.layer2.0.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.21 : Tensor = aten::batch_norm(%input.155, %self.kp_detector.fg_encoder.layer2.0.bn2.weight, %self.kp_detector.fg_encoder.layer2.0.bn2.bias, %self.kp_detector.fg_encoder.layer2.0.bn2.running_mean, %self.kp_detector.fg_encoder.layer2.0.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.0/__module.kp_detector.fg_encoder.layer2.0.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 Set kernel index: 3
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Finalize: %input.271 : Tensor = aten::_convolution(%310, %self.kp_detector.fg_encoder.layer2.0.downsample.0.weight, %5, %25, %28, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.0/__module.kp_detector.fg_encoder.layer2.0.downsample/__module.kp_detector.fg_encoder.layer2.0.downsample.0 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %identity.13 : Tensor = aten::batch_norm(%input.271, %self.kp_detector.fg_encoder.layer2.0.downsample.1.weight, %self.kp_detector.fg_encoder.layer2.0.bn2.bias, %self.kp_detector.fg_encoder.layer2.0.downsample.1.running_mean, %self.kp_detector.fg_encoder.layer2.0.downsample.1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.0/__module.kp_detector.fg_encoder.layer2.0.downsample/__module.kp_detector.fg_encoder.layer2.0.downsample.1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %318 : Tensor = aten::add(%out.37, %identity.13, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.0 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %319 : Tensor = aten::relu(%318), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.0/__module.kp_detector.fg_encoder.layer2.0.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 Set kernel index: 4
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Finalize: %input.157 : Tensor = aten::_convolution(%231, %self.kp_detector.fg_encoder.layer2.0.downsample.0.weight, %5, %25, %28, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.0/__module.kp_detector.fg_encoder.layer2.0.downsample/__module.kp_detector.fg_encoder.layer2.0.downsample.0 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %identity.7 : Tensor = aten::batch_norm(%input.157, %self.kp_detector.fg_encoder.layer2.0.downsample.1.weight, %self.kp_detector.fg_encoder.layer2.0.bn2.bias, %self.kp_detector.fg_encoder.layer2.0.downsample.1.running_mean, %self.kp_detector.fg_encoder.layer2.0.downsample.1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.0/__module.kp_detector.fg_encoder.layer2.0.downsample/__module.kp_detector.fg_encoder.layer2.0.downsample.1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %239 : Tensor = aten::add(%out.21, %identity.7, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.0 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %240 : Tensor = aten::relu(%239), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.0/__module.kp_detector.fg_encoder.layer2.0.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 Set kernel index: 4
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Finalize: %input.163 : Tensor = aten::_convolution(%240, %self.kp_detector.fg_encoder.layer2.1.conv1.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.1/__module.kp_detector.fg_encoder.layer2.1.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.165 : Tensor = aten::batch_norm(%input.163, %self.kp_detector.fg_encoder.layer2.1.bn1.weight, %self.kp_detector.fg_encoder.layer2.1.bn1.bias, %self.kp_detector.fg_encoder.layer2.1.bn1.running_mean, %self.kp_detector.fg_encoder.layer2.1.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.1/__module.kp_detector.fg_encoder.layer2.1.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %243 : Tensor = aten::relu(%input.165), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.1/__module.kp_detector.fg_encoder.layer2.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 Set kernel index: 3
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Finalize: %input.277 : Tensor = aten::_convolution(%319, %self.kp_detector.fg_encoder.layer2.1.conv1.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.1/__module.kp_detector.fg_encoder.layer2.1.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.279 : Tensor = aten::batch_norm(%input.277, %self.kp_detector.fg_encoder.layer2.1.bn1.weight, %self.kp_detector.fg_encoder.layer2.1.bn1.bias, %self.kp_detector.fg_encoder.layer2.1.bn1.running_mean, %self.kp_detector.fg_encoder.layer2.1.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.1/__module.kp_detector.fg_encoder.layer2.1.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %322 : Tensor = aten::relu(%input.279), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.1/__module.kp_detector.fg_encoder.layer2.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 Set kernel index: 3
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Finalize: %input.283 : Tensor = aten::_convolution(%322, %self.kp_detector.fg_encoder.layer2.1.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.1/__module.kp_detector.fg_encoder.layer2.1.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.39 : Tensor = aten::batch_norm(%input.283, %self.kp_detector.fg_encoder.layer2.1.bn2.weight, %self.kp_detector.fg_encoder.layer2.1.bn2.bias, %self.kp_detector.fg_encoder.layer2.1.bn2.running_mean, %self.kp_detector.fg_encoder.layer2.1.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.1/__module.kp_detector.fg_encoder.layer2.1.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %325 : Tensor = aten::add(%out.39, %319, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.1 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %326 : Tensor = aten::relu(%325), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.1/__module.kp_detector.fg_encoder.layer2.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 Set kernel index: 3
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Finalize: %input.169 : Tensor = aten::_convolution(%243, %self.kp_detector.fg_encoder.layer2.1.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.1/__module.kp_detector.fg_encoder.layer2.1.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.23 : Tensor = aten::batch_norm(%input.169, %self.kp_detector.fg_encoder.layer2.1.bn2.weight, %self.kp_detector.fg_encoder.layer2.1.bn2.bias, %self.kp_detector.fg_encoder.layer2.1.bn2.running_mean, %self.kp_detector.fg_encoder.layer2.1.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.1/__module.kp_detector.fg_encoder.layer2.1.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %246 : Tensor = aten::add(%out.23, %240, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.1 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %247 : Tensor = aten::relu(%246), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.1/__module.kp_detector.fg_encoder.layer2.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 Set kernel index: 3
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Finalize: %input.175 : Tensor = aten::_convolution(%247, %self.kp_detector.fg_encoder.layer3.0.conv1.weight, %5, %25, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.177 : Tensor = aten::batch_norm(%input.175, %self.kp_detector.fg_encoder.layer3.0.bn1.weight, %self.kp_detector.fg_encoder.layer3.0.bn1.bias, %self.kp_detector.fg_encoder.layer3.0.bn1.running_mean, %self.kp_detector.fg_encoder.layer3.0.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %250 : Tensor = aten::relu(%input.177), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 Set kernel index: 5
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Finalize: %input.289 : Tensor = aten::_convolution(%326, %self.kp_detector.fg_encoder.layer3.0.conv1.weight, %5, %25, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.291 : Tensor = aten::batch_norm(%input.289, %self.kp_detector.fg_encoder.layer3.0.bn1.weight, %self.kp_detector.fg_encoder.layer3.0.bn1.bias, %self.kp_detector.fg_encoder.layer3.0.bn1.running_mean, %self.kp_detector.fg_encoder.layer3.0.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %329 : Tensor = aten::relu(%input.291), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 Set kernel index: 5
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Finalize: %input.295 : Tensor = aten::_convolution(%329, %self.kp_detector.fg_encoder.layer3.0.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.41 : Tensor = aten::batch_norm(%input.295, %self.kp_detector.fg_encoder.layer3.0.bn2.weight, %self.kp_detector.fg_encoder.layer3.0.bn2.bias, %self.kp_detector.fg_encoder.layer3.0.bn2.running_mean, %self.kp_detector.fg_encoder.layer3.0.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 Set kernel index: 5
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Finalize: %input.181 : Tensor = aten::_convolution(%250, %self.kp_detector.fg_encoder.layer3.0.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.25 : Tensor = aten::batch_norm(%input.181, %self.kp_detector.fg_encoder.layer3.0.bn2.weight, %self.kp_detector.fg_encoder.layer3.0.bn2.bias, %self.kp_detector.fg_encoder.layer3.0.bn2.running_mean, %self.kp_detector.fg_encoder.layer3.0.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 Set kernel index: 5
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Finalize: %input.297 : Tensor = aten::_convolution(%326, %self.kp_detector.fg_encoder.layer3.0.downsample.0.weight, %5, %25, %28, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.downsample/__module.kp_detector.fg_encoder.layer3.0.downsample.0 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %identity.15 : Tensor = aten::batch_norm(%input.297, %self.kp_detector.fg_encoder.layer3.0.downsample.1.weight, %self.kp_detector.fg_encoder.layer3.0.bn2.bias, %self.kp_detector.fg_encoder.layer3.0.downsample.1.running_mean, %self.kp_detector.fg_encoder.layer3.0.downsample.1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.downsample/__module.kp_detector.fg_encoder.layer3.0.downsample.1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %334 : Tensor = aten::add(%out.41, %identity.15, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %335 : Tensor = aten::relu(%334), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 Set kernel index: 4
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Finalize: %input.183 : Tensor = aten::_convolution(%247, %self.kp_detector.fg_encoder.layer3.0.downsample.0.weight, %5, %25, %28, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.downsample/__module.kp_detector.fg_encoder.layer3.0.downsample.0 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %identity.9 : Tensor = aten::batch_norm(%input.183, %self.kp_detector.fg_encoder.layer3.0.downsample.1.weight, %self.kp_detector.fg_encoder.layer3.0.bn2.bias, %self.kp_detector.fg_encoder.layer3.0.downsample.1.running_mean, %self.kp_detector.fg_encoder.layer3.0.downsample.1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.downsample/__module.kp_detector.fg_encoder.layer3.0.downsample.1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %255 : Tensor = aten::add(%out.25, %identity.9, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %256 : Tensor = aten::relu(%255), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 Set kernel index: 4
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Finalize: %input.189 : Tensor = aten::_convolution(%256, %self.kp_detector.fg_encoder.layer3.1.conv1.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1/__module.kp_detector.fg_encoder.layer3.1.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.191 : Tensor = aten::batch_norm(%input.189, %self.kp_detector.fg_encoder.layer3.1.bn1.weight, %self.kp_detector.fg_encoder.layer3.1.bn1.bias, %self.kp_detector.fg_encoder.layer3.1.bn1.running_mean, %self.kp_detector.fg_encoder.layer3.1.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1/__module.kp_detector.fg_encoder.layer3.1.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %259 : Tensor = aten::relu(%input.191), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1/__module.kp_detector.fg_encoder.layer3.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 Set kernel index: 5
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Finalize: %input.303 : Tensor = aten::_convolution(%335, %self.kp_detector.fg_encoder.layer3.1.conv1.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1/__module.kp_detector.fg_encoder.layer3.1.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.305 : Tensor = aten::batch_norm(%input.303, %self.kp_detector.fg_encoder.layer3.1.bn1.weight, %self.kp_detector.fg_encoder.layer3.1.bn1.bias, %self.kp_detector.fg_encoder.layer3.1.bn1.running_mean, %self.kp_detector.fg_encoder.layer3.1.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1/__module.kp_detector.fg_encoder.layer3.1.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %338 : Tensor = aten::relu(%input.305), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1/__module.kp_detector.fg_encoder.layer3.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 Set kernel index: 5
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Finalize: %input.309 : Tensor = aten::_convolution(%338, %self.kp_detector.fg_encoder.layer3.1.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1/__module.kp_detector.fg_encoder.layer3.1.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.43 : Tensor = aten::batch_norm(%input.309, %self.kp_detector.fg_encoder.layer3.1.bn2.weight, %self.kp_detector.fg_encoder.layer3.1.bn2.bias, %self.kp_detector.fg_encoder.layer3.1.bn2.running_mean, %self.kp_detector.fg_encoder.layer3.1.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1/__module.kp_detector.fg_encoder.layer3.1.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %341 : Tensor = aten::add(%out.43, %335, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %342 : Tensor = aten::relu(%341), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1/__module.kp_detector.fg_encoder.layer3.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 Set kernel index: 5
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Finalize: %input.195 : Tensor = aten::_convolution(%259, %self.kp_detector.fg_encoder.layer3.1.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1/__module.kp_detector.fg_encoder.layer3.1.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.27 : Tensor = aten::batch_norm(%input.195, %self.kp_detector.fg_encoder.layer3.1.bn2.weight, %self.kp_detector.fg_encoder.layer3.1.bn2.bias, %self.kp_detector.fg_encoder.layer3.1.bn2.running_mean, %self.kp_detector.fg_encoder.layer3.1.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1/__module.kp_detector.fg_encoder.layer3.1.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %262 : Tensor = aten::add(%out.27, %256, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %263 : Tensor = aten::relu(%262), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1/__module.kp_detector.fg_encoder.layer3.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 Set kernel index: 5
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Finalize: %input.201 : Tensor = aten::_convolution(%263, %self.kp_detector.fg_encoder.layer4.0.conv1.weight, %5, %25, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.203 : Tensor = aten::batch_norm(%input.201, %self.kp_detector.fg_encoder.layer4.0.bn1.weight, %self.kp_detector.fg_encoder.layer4.0.bn1.bias, %self.kp_detector.fg_encoder.layer4.0.bn1.running_mean, %self.kp_detector.fg_encoder.layer4.0.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %266 : Tensor = aten::relu(%input.203), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 Set kernel index: 5
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Finalize: %input.315 : Tensor = aten::_convolution(%342, %self.kp_detector.fg_encoder.layer4.0.conv1.weight, %5, %25, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.317 : Tensor = aten::batch_norm(%input.315, %self.kp_detector.fg_encoder.layer4.0.bn1.weight, %self.kp_detector.fg_encoder.layer4.0.bn1.bias, %self.kp_detector.fg_encoder.layer4.0.bn1.running_mean, %self.kp_detector.fg_encoder.layer4.0.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %345 : Tensor = aten::relu(%input.317), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 Set kernel index: 5
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Finalize: %input.321 : Tensor = aten::_convolution(%345, %self.kp_detector.fg_encoder.layer4.0.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.45 : Tensor = aten::batch_norm(%input.321, %self.kp_detector.fg_encoder.layer4.0.bn2.weight, %self.kp_detector.fg_encoder.layer4.0.bn2.bias, %self.kp_detector.fg_encoder.layer4.0.bn2.running_mean, %self.kp_detector.fg_encoder.layer4.0.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 Set kernel index: 5
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Finalize: %input.207 : Tensor = aten::_convolution(%266, %self.kp_detector.fg_encoder.layer4.0.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.29 : Tensor = aten::batch_norm(%input.207, %self.kp_detector.fg_encoder.layer4.0.bn2.weight, %self.kp_detector.fg_encoder.layer4.0.bn2.bias, %self.kp_detector.fg_encoder.layer4.0.bn2.running_mean, %self.kp_detector.fg_encoder.layer4.0.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 Set kernel index: 5
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Finalize: %input.323 : Tensor = aten::_convolution(%342, %self.kp_detector.fg_encoder.layer4.0.downsample.0.weight, %5, %25, %28, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.downsample/__module.kp_detector.fg_encoder.layer4.0.downsample.0 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %identity : Tensor = aten::batch_norm(%input.323, %self.kp_detector.fg_encoder.layer4.0.downsample.1.weight, %self.kp_detector.fg_encoder.layer4.0.bn2.bias, %self.kp_detector.fg_encoder.layer4.0.downsample.1.running_mean, %self.kp_detector.fg_encoder.layer4.0.downsample.1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.downsample/__module.kp_detector.fg_encoder.layer4.0.downsample.1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %350 : Tensor = aten::add(%out.45, %identity, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %351 : Tensor = aten::relu(%350), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 Set kernel index: 4
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Finalize: %input.209 : Tensor = aten::_convolution(%263, %self.kp_detector.fg_encoder.layer4.0.downsample.0.weight, %5, %25, %28, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.downsample/__module.kp_detector.fg_encoder.layer4.0.downsample.0 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %identity.11 : Tensor = aten::batch_norm(%input.209, %self.kp_detector.fg_encoder.layer4.0.downsample.1.weight, %self.kp_detector.fg_encoder.layer4.0.bn2.bias, %self.kp_detector.fg_encoder.layer4.0.downsample.1.running_mean, %self.kp_detector.fg_encoder.layer4.0.downsample.1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.downsample/__module.kp_detector.fg_encoder.layer4.0.downsample.1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %271 : Tensor = aten::add(%out.29, %identity.11, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %272 : Tensor = aten::relu(%271), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 Set kernel index: 4
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Finalize: %input.215 : Tensor = aten::_convolution(%272, %self.kp_detector.fg_encoder.layer4.1.conv1.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1/__module.kp_detector.fg_encoder.layer4.1.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.217 : Tensor = aten::batch_norm(%input.215, %self.kp_detector.fg_encoder.layer4.1.bn1.weight, %self.kp_detector.fg_encoder.layer4.1.bn1.bias, %self.kp_detector.fg_encoder.layer4.1.bn1.running_mean, %self.kp_detector.fg_encoder.layer4.1.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1/__module.kp_detector.fg_encoder.layer4.1.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %275 : Tensor = aten::relu(%input.217), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1/__module.kp_detector.fg_encoder.layer4.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 Set kernel index: 5
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Finalize: %input.329 : Tensor = aten::_convolution(%351, %self.kp_detector.fg_encoder.layer4.1.conv1.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1/__module.kp_detector.fg_encoder.layer4.1.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.331 : Tensor = aten::batch_norm(%input.329, %self.kp_detector.fg_encoder.layer4.1.bn1.weight, %self.kp_detector.fg_encoder.layer4.1.bn1.bias, %self.kp_detector.fg_encoder.layer4.1.bn1.running_mean, %self.kp_detector.fg_encoder.layer4.1.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1/__module.kp_detector.fg_encoder.layer4.1.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %354 : Tensor = aten::relu(%input.331), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1/__module.kp_detector.fg_encoder.layer4.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 Set kernel index: 5
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Finalize: %input.335 : Tensor = aten::_convolution(%354, %self.kp_detector.fg_encoder.layer4.1.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1/__module.kp_detector.fg_encoder.layer4.1.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.47 : Tensor = aten::batch_norm(%input.335, %self.kp_detector.fg_encoder.layer4.1.bn2.weight, %self.kp_detector.fg_encoder.layer4.1.bn2.bias, %self.kp_detector.fg_encoder.layer4.1.bn2.running_mean, %self.kp_detector.fg_encoder.layer4.1.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1/__module.kp_detector.fg_encoder.layer4.1.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %357 : Tensor = aten::add(%out.47, %351, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %358 : Tensor = aten::relu(%357), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1/__module.kp_detector.fg_encoder.layer4.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 Set kernel index: 5
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Finalize: %x.9 : Tensor = aten::adaptive_avg_pool2d(%358, %27), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.avgpool # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1213:0 Set kernel index: 6
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Finalize: %input.221 : Tensor = aten::_convolution(%275, %self.kp_detector.fg_encoder.layer4.1.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1/__module.kp_detector.fg_encoder.layer4.1.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.31 : Tensor = aten::batch_norm(%input.221, %self.kp_detector.fg_encoder.layer4.1.bn2.weight, %self.kp_detector.fg_encoder.layer4.1.bn2.bias, %self.kp_detector.fg_encoder.layer4.1.bn2.running_mean, %self.kp_detector.fg_encoder.layer4.1.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1/__module.kp_detector.fg_encoder.layer4.1.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %278 : Tensor = aten::add(%out.31, %272, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %279 : Tensor = aten::relu(%278), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1/__module.kp_detector.fg_encoder.layer4.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 Set kernel index: 5
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Finalize: %x.7 : Tensor = aten::adaptive_avg_pool2d(%279, %27), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.avgpool # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1213:0 Set kernel index: 6
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Finalize: PWN(%mean_diff : Tensor = aten::sub(%372, %374, %365) # /home/joy/remote/stable-diffusion-backend/personalize/thin_plate_spline_motion_model/tpsnn.py:503:0, %376 : Tensor = aten::pow(%mean_diff, %21) # /home/joy/remote/stable-diffusion-backend/personalize/thin_plate_spline_motion_model/tpsnn.py:503:0) Set kernel index: 9
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Finalize: PWN(PWN(%kp_value_diff.1 : Tensor = aten::sub(%371, %292, %365) # /home/joy/remote/stable-diffusion-backend/personalize/thin_plate_spline_motion_model/tpsnn.py:510:0, %380 : Tensor = aten::mul(%kp_value_diff.1, %adapt_movement_scale) # /home/joy/remote/stable-diffusion-backend/personalize/thin_plate_spline_motion_model/tpsnn.py:511:0), %kp.1 : Tensor = aten::add(%380, %211, %365) # /home/joy/remote/stable-diffusion-backend/personalize/thin_plate_spline_motion_model/tpsnn.py:511:0) Set kernel index: 10
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Total number of generated kernels selected for the engine: 11
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Kernel: 0 CASK_STATIC
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Kernel: 1 CASK_STATIC
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Kernel: 2 CASK_STATIC
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Kernel: 3 CASK_STATIC
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Kernel: 4 CASK_STATIC
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Kernel: 5 CASK_STATIC
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Kernel: 6 CASK_STATIC
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Kernel: 7 CASK_STATIC
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Kernel: 8 CASK_STATIC
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Kernel: 9 TRT_SERIALIZABLE:generatedNativePointwise
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Kernel: 10 TRT_SERIALIZABLE:generatedNativePointwise
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Disabling unused tactic source: CUDNN
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Disabling unused tactic source: JIT_CONVOLUTIONS
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Trying to load shared library libcublas.so.11
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Loaded shared library libcublas.so.11
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Using cublas as plugin tactic source
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Trying to load shared library libcublasLt.so.11
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Loaded shared library libcublasLt.so.11
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Using cublasLt as core library tactic source
INFO: [Torch-TensorRT TorchScript Conversion Context] - [MemUsageChange] Init cuBLAS/cuBLASLt: CPU +1, GPU +8, now: CPU 4482, GPU 13857 (MiB)
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Engine generation completed in 35.5369 seconds.
WARNING: [Torch-TensorRT TorchScript Conversion Context] - TensorRT encountered issues when converting weights between types and that could affect accuracy.
WARNING: [Torch-TensorRT TorchScript Conversion Context] - If this is not the desired behavior, please modify the weights or retrain with regularization to adjust the magnitude of the weights.
WARNING: [Torch-TensorRT TorchScript Conversion Context] - Check verbose logs for the list of affected weights.
WARNING: [Torch-TensorRT TorchScript Conversion Context] - - 63 weights are affected by this issue: Detected subnormal FP16 values.
DEBUG: [Torch-TensorRT TorchScript Conversion Context] -   List of affected weights: %198 : Tensor = aten::matmul(%input.113, %196) + [Freeze Tensor %199 : Tensor = trt::const(%self.kp_detector.fg_encoder.fc.bias) ] + (Unnamed Layer* 83) [Shuffle] + unsqueeze_node_after_[Freeze Tensor %199 : Tensor = trt::const(%self.kp_detector.fg_encoder.fc.bias) ] + (Unnamed Layer* 83) [Shuffle]_(Unnamed Layer* 83) [Shuffle]_output + %201 : Tensor = aten::add(%199, %198, %365) + PWN(%fg_kp.3 : Tensor = aten::sigmoid(%201), scope: __module.kp_detector # /home/joy/remote/stable-diffusion-backend/personalize/thin_plate_spline_motion_model/tpsnn.py:43:0).weight, %283 : Tensor = aten::matmul(%input.227, %282) + [Freeze Tensor %284 : Tensor = trt::const(%self.kp_detector.fg_encoder.fc.bias) ] + (Unnamed Layer* 168) [Shuffle] + unsqueeze_node_after_[Freeze Tensor %284 : Tensor = trt::const(%self.kp_detector.fg_encoder.fc.bias) ] + (Unnamed Layer* 168) [Shuffle]_(Unnamed Layer* 168) [Shuffle]_output + %285 : Tensor = aten::add(%284, %283, %365) + PWN(%fg_kp.9 : Tensor = aten::sigmoid(%285), scope: __module.kp_detector # /home/joy/remote/stable-diffusion-backend/personalize/thin_plate_spline_motion_model/tpsnn.py:43:0).weight, %362 : Tensor = aten::matmul(%input.341, %361) + [Freeze Tensor %363 : Tensor = trt::const(%self.kp_detector.fg_encoder.fc.bias) ] + (Unnamed Layer* 252) [Shuffle] + unsqueeze_node_after_[Freeze Tensor %363 : Tensor = trt::const(%self.kp_detector.fg_encoder.fc.bias) ] + (Unnamed Layer* 252) [Shuffle]_(Unnamed Layer* 252) [Shuffle]_output + %364 : Tensor = aten::add(%363, %362, %365) + PWN(%fg_kp.15 : Tensor = aten::sigmoid(%364), scope: __module.kp_detector # /home/joy/remote/stable-diffusion-backend/personalize/thin_plate_spline_motion_model/tpsnn.py:43:0).weight, %input.101 : Tensor = aten::_convolution(%175, %self.kp_detector.fg_encoder.layer4.1.conv1.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1/__module.kp_detector.fg_encoder.layer4.1.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.103 : Tensor = aten::batch_norm(%input.101, %self.kp_detector.fg_encoder.layer4.1.bn1.weight, %self.kp_detector.fg_encoder.layer4.1.bn1.bias, %self.kp_detector.fg_encoder.layer4.1.bn1.running_mean, %self.kp_detector.fg_encoder.layer4.1.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1/__module.kp_detector.fg_encoder.layer4.1.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %183 : Tensor = aten::relu(%input.103), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1/__module.kp_detector.fg_encoder.layer4.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0.weight, %input.107 : Tensor = aten::_convolution(%183, %self.kp_detector.fg_encoder.layer4.1.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1/__module.kp_detector.fg_encoder.layer4.1.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.15 : Tensor = aten::batch_norm(%input.107, %self.kp_detector.fg_encoder.layer4.1.bn2.weight, %self.kp_detector.fg_encoder.layer4.1.bn2.bias, %self.kp_detector.fg_encoder.layer4.1.bn2.running_mean, %self.kp_detector.fg_encoder.layer4.1.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1/__module.kp_detector.fg_encoder.layer4.1.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %191 : Tensor = aten::add(%out.15, %175, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %192 : Tensor = aten::relu(%191), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1/__module.kp_detector.fg_encoder.layer4.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0.weight, %input.11 : Tensor = aten::_convolution(%input.9, %self.kp_detector.fg_encoder.layer1.0.conv1.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.0/__module.kp_detector.fg_encoder.layer1.0.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.13 : Tensor = aten::batch_norm(%input.11, %self.kp_detector.fg_encoder.layer1.0.bn1.weight, %self.kp_detector.fg_encoder.layer1.0.bn1.bias, %self.kp_detector.fg_encoder.layer1.0.bn1.running_mean, %self.kp_detector.fg_encoder.layer1.0.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.0/__module.kp_detector.fg_encoder.layer1.0.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %46 : Tensor = aten::relu(%input.13), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.0/__module.kp_detector.fg_encoder.layer1.0.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0.weight, %input.125 : Tensor = aten::_convolution(%input.123, %self.kp_detector.fg_encoder.layer1.0.conv1.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.0/__module.kp_detector.fg_encoder.layer1.0.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.127 : Tensor = aten::batch_norm(%input.125, %self.kp_detector.fg_encoder.layer1.0.bn1.weight, %self.kp_detector.fg_encoder.layer1.0.bn1.bias, %self.kp_detector.fg_encoder.layer1.0.bn1.running_mean, %self.kp_detector.fg_encoder.layer1.0.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.0/__module.kp_detector.fg_encoder.layer1.0.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %220 : Tensor = aten::relu(%input.127), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.0/__module.kp_detector.fg_encoder.layer1.0.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0.weight, %input.131 : Tensor = aten::_convolution(%220, %self.kp_detector.fg_encoder.layer1.0.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.0/__module.kp_detector.fg_encoder.layer1.0.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.17 : Tensor = aten::batch_norm(%input.131, %self.kp_detector.fg_encoder.layer1.0.bn2.weight, %self.kp_detector.fg_encoder.layer1.0.bn2.bias, %self.kp_detector.fg_encoder.layer1.0.bn2.running_mean, %self.kp_detector.fg_encoder.layer1.0.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.0/__module.kp_detector.fg_encoder.layer1.0.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %223 : Tensor = aten::add(%out.17, %input.123, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.0 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %224 : Tensor = aten::relu(%223), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.0/__module.kp_detector.fg_encoder.layer1.0.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0.weight, %input.137 : Tensor = aten::_convolution(%224, %self.kp_detector.fg_encoder.layer1.1.conv1.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.1/__module.kp_detector.fg_encoder.layer1.1.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.139 : Tensor = aten::batch_norm(%input.137, %self.kp_detector.fg_encoder.layer1.1.bn1.weight, %self.kp_detector.fg_encoder.layer1.1.bn1.bias, %self.kp_detector.fg_encoder.layer1.1.bn1.running_mean, %self.kp_detector.fg_encoder.layer1.1.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.1/__module.kp_detector.fg_encoder.layer1.1.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %227 : Tensor = aten::relu(%input.139), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.1/__module.kp_detector.fg_encoder.layer1.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0.weight, %input.143 : Tensor = aten::_convolution(%227, %self.kp_detector.fg_encoder.layer1.1.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.1/__module.kp_detector.fg_encoder.layer1.1.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.19 : Tensor = aten::batch_norm(%input.143, %self.kp_detector.fg_encoder.layer1.1.bn2.weight, %self.kp_detector.fg_encoder.layer1.1.bn2.bias, %self.kp_detector.fg_encoder.layer1.1.bn2.running_mean, %self.kp_detector.fg_encoder.layer1.1.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.1/__module.kp_detector.fg_encoder.layer1.1.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %230 : Tensor = aten::add(%out.19, %224, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.1 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %231 : Tensor = aten::relu(%230), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.1/__module.kp_detector.fg_encoder.layer1.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0.weight, %input.149 : Tensor = aten::_convolution(%231, %self.kp_detector.fg_encoder.layer2.0.conv1.weight, %5, %25, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.0/__module.kp_detector.fg_encoder.layer2.0.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.151 : Tensor = aten::batch_norm(%input.149, %self.kp_detector.fg_encoder.layer2.0.bn1.weight, %self.kp_detector.fg_encoder.layer2.0.bn1.bias, %self.kp_detector.fg_encoder.layer2.0.bn1.running_mean, %self.kp_detector.fg_encoder.layer2.0.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.0/__module.kp_detector.fg_encoder.layer2.0.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %234 : Tensor = aten::relu(%input.151), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.0/__module.kp_detector.fg_encoder.layer2.0.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0.weight, %input.155 : Tensor = aten::_convolution(%234, %self.kp_detector.fg_encoder.layer2.0.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.0/__module.kp_detector.fg_encoder.layer2.0.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.21 : Tensor = aten::batch_norm(%input.155, %self.kp_detector.fg_encoder.layer2.0.bn2.weight, %self.kp_detector.fg_encoder.layer2.0.bn2.bias, %self.kp_detector.fg_encoder.layer2.0.bn2.running_mean, %self.kp_detector.fg_encoder.layer2.0.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.0/__module.kp_detector.fg_encoder.layer2.0.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0.weight, %input.157 : Tensor = aten::_convolution(%231, %self.kp_detector.fg_encoder.layer2.0.downsample.0.weight, %5, %25, %28, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.0/__module.kp_detector.fg_encoder.layer2.0.downsample/__module.kp_detector.fg_encoder.layer2.0.downsample.0 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %identity.7 : Tensor = aten::batch_norm(%input.157, %self.kp_detector.fg_encoder.layer2.0.downsample.1.weight, %self.kp_detector.fg_encoder.layer2.0.bn2.bias, %self.kp_detector.fg_encoder.layer2.0.downsample.1.running_mean, %self.kp_detector.fg_encoder.layer2.0.downsample.1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.0/__module.kp_detector.fg_encoder.layer2.0.downsample/__module.kp_detector.fg_encoder.layer2.0.downsample.1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %239 : Tensor = aten::add(%out.21, %identity.7, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.0 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %240 : Tensor = aten::relu(%239), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.0/__module.kp_detector.fg_encoder.layer2.0.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0.bias, %input.157 : Tensor = aten::_convolution(%231, %self.kp_detector.fg_encoder.layer2.0.downsample.0.weight, %5, %25, %28, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.0/__module.kp_detector.fg_encoder.layer2.0.downsample/__module.kp_detector.fg_encoder.layer2.0.downsample.0 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %identity.7 : Tensor = aten::batch_norm(%input.157, %self.kp_detector.fg_encoder.layer2.0.downsample.1.weight, %self.kp_detector.fg_encoder.layer2.0.bn2.bias, %self.kp_detector.fg_encoder.layer2.0.downsample.1.running_mean, %self.kp_detector.fg_encoder.layer2.0.downsample.1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.0/__module.kp_detector.fg_encoder.layer2.0.downsample/__module.kp_detector.fg_encoder.layer2.0.downsample.1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %239 : Tensor = aten::add(%out.21, %identity.7, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.0 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %240 : Tensor = aten::relu(%239), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.0/__module.kp_detector.fg_encoder.layer2.0.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0.weight, %input.163 : Tensor = aten::_convolution(%240, %self.kp_detector.fg_encoder.layer2.1.conv1.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.1/__module.kp_detector.fg_encoder.layer2.1.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.165 : Tensor = aten::batch_norm(%input.163, %self.kp_detector.fg_encoder.layer2.1.bn1.weight, %self.kp_detector.fg_encoder.layer2.1.bn1.bias, %self.kp_detector.fg_encoder.layer2.1.bn1.running_mean, %self.kp_detector.fg_encoder.layer2.1.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.1/__module.kp_detector.fg_encoder.layer2.1.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %243 : Tensor = aten::relu(%input.165), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.1/__module.kp_detector.fg_encoder.layer2.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0.weight, %input.169 : Tensor = aten::_convolution(%243, %self.kp_detector.fg_encoder.layer2.1.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.1/__module.kp_detector.fg_encoder.layer2.1.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.23 : Tensor = aten::batch_norm(%input.169, %self.kp_detector.fg_encoder.layer2.1.bn2.weight, %self.kp_detector.fg_encoder.layer2.1.bn2.bias, %self.kp_detector.fg_encoder.layer2.1.bn2.running_mean, %self.kp_detector.fg_encoder.layer2.1.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.1/__module.kp_detector.fg_encoder.layer2.1.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %246 : Tensor = aten::add(%out.23, %240, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.1 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %247 : Tensor = aten::relu(%246), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.1/__module.kp_detector.fg_encoder.layer2.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0.weight, %input.17 : Tensor = aten::_convolution(%46, %self.kp_detector.fg_encoder.layer1.0.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.0/__module.kp_detector.fg_encoder.layer1.0.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.1 : Tensor = aten::batch_norm(%input.17, %self.kp_detector.fg_encoder.layer1.0.bn2.weight, %self.kp_detector.fg_encoder.layer1.0.bn2.bias, %self.kp_detector.fg_encoder.layer1.0.bn2.running_mean, %self.kp_detector.fg_encoder.layer1.0.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.0/__module.kp_detector.fg_encoder.layer1.0.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %54 : Tensor = aten::add(%out.1, %input.9, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.0 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %55 : Tensor = aten::relu(%54), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.0/__module.kp_detector.fg_encoder.layer1.0.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0.weight, %input.175 : Tensor = aten::_convolution(%247, %self.kp_detector.fg_encoder.layer3.0.conv1.weight, %5, %25, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.177 : Tensor = aten::batch_norm(%input.175, %self.kp_detector.fg_encoder.layer3.0.bn1.weight, %self.kp_detector.fg_encoder.layer3.0.bn1.bias, %self.kp_detector.fg_encoder.layer3.0.bn1.running_mean, %self.kp_detector.fg_encoder.layer3.0.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %250 : Tensor = aten::relu(%input.177), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0.weight, %input.181 : Tensor = aten::_convolution(%250, %self.kp_detector.fg_encoder.layer3.0.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.25 : Tensor = aten::batch_norm(%input.181, %self.kp_detector.fg_encoder.layer3.0.bn2.weight, %self.kp_detector.fg_encoder.layer3.0.bn2.bias, %self.kp_detector.fg_encoder.layer3.0.bn2.running_mean, %self.kp_detector.fg_encoder.layer3.0.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0.weight, %input.183 : Tensor = aten::_convolution(%247, %self.kp_detector.fg_encoder.layer3.0.downsample.0.weight, %5, %25, %28, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.downsample/__module.kp_detector.fg_encoder.layer3.0.downsample.0 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %identity.9 : Tensor = aten::batch_norm(%input.183, %self.kp_detector.fg_encoder.layer3.0.downsample.1.weight, %self.kp_detector.fg_encoder.layer3.0.bn2.bias, %self.kp_detector.fg_encoder.layer3.0.downsample.1.running_mean, %self.kp_detector.fg_encoder.layer3.0.downsample.1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.downsample/__module.kp_detector.fg_encoder.layer3.0.downsample.1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %255 : Tensor = aten::add(%out.25, %identity.9, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %256 : Tensor = aten::relu(%255), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0.weight, %input.189 : Tensor = aten::_convolution(%256, %self.kp_detector.fg_encoder.layer3.1.conv1.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1/__module.kp_detector.fg_encoder.layer3.1.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.191 : Tensor = aten::batch_norm(%input.189, %self.kp_detector.fg_encoder.layer3.1.bn1.weight, %self.kp_detector.fg_encoder.layer3.1.bn1.bias, %self.kp_detector.fg_encoder.layer3.1.bn1.running_mean, %self.kp_detector.fg_encoder.layer3.1.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1/__module.kp_detector.fg_encoder.layer3.1.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %259 : Tensor = aten::relu(%input.191), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1/__module.kp_detector.fg_encoder.layer3.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0.weight, %input.195 : Tensor = aten::_convolution(%259, %self.kp_detector.fg_encoder.layer3.1.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1/__module.kp_detector.fg_encoder.layer3.1.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.27 : Tensor = aten::batch_norm(%input.195, %self.kp_detector.fg_encoder.layer3.1.bn2.weight, %self.kp_detector.fg_encoder.layer3.1.bn2.bias, %self.kp_detector.fg_encoder.layer3.1.bn2.running_mean, %self.kp_detector.fg_encoder.layer3.1.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1/__module.kp_detector.fg_encoder.layer3.1.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %262 : Tensor = aten::add(%out.27, %256, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %263 : Tensor = aten::relu(%262), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1/__module.kp_detector.fg_encoder.layer3.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0.weight, %input.201 : Tensor = aten::_convolution(%263, %self.kp_detector.fg_encoder.layer4.0.conv1.weight, %5, %25, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.203 : Tensor = aten::batch_norm(%input.201, %self.kp_detector.fg_encoder.layer4.0.bn1.weight, %self.kp_detector.fg_encoder.layer4.0.bn1.bias, %self.kp_detector.fg_encoder.layer4.0.bn1.running_mean, %self.kp_detector.fg_encoder.layer4.0.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %266 : Tensor = aten::relu(%input.203), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0.weight, %input.207 : Tensor = aten::_convolution(%266, %self.kp_detector.fg_encoder.layer4.0.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.29 : Tensor = aten::batch_norm(%input.207, %self.kp_detector.fg_encoder.layer4.0.bn2.weight, %self.kp_detector.fg_encoder.layer4.0.bn2.bias, %self.kp_detector.fg_encoder.layer4.0.bn2.running_mean, %self.kp_detector.fg_encoder.layer4.0.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0.weight, %input.209 : Tensor = aten::_convolution(%263, %self.kp_detector.fg_encoder.layer4.0.downsample.0.weight, %5, %25, %28, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.downsample/__module.kp_detector.fg_encoder.layer4.0.downsample.0 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %identity.11 : Tensor = aten::batch_norm(%input.209, %self.kp_detector.fg_encoder.layer4.0.downsample.1.weight, %self.kp_detector.fg_encoder.layer4.0.bn2.bias, %self.kp_detector.fg_encoder.layer4.0.downsample.1.running_mean, %self.kp_detector.fg_encoder.layer4.0.downsample.1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.downsample/__module.kp_detector.fg_encoder.layer4.0.downsample.1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %271 : Tensor = aten::add(%out.29, %identity.11, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %272 : Tensor = aten::relu(%271), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0.weight, %input.215 : Tensor = aten::_convolution(%272, %self.kp_detector.fg_encoder.layer4.1.conv1.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1/__module.kp_detector.fg_encoder.layer4.1.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.217 : Tensor = aten::batch_norm(%input.215, %self.kp_detector.fg_encoder.layer4.1.bn1.weight, %self.kp_detector.fg_encoder.layer4.1.bn1.bias, %self.kp_detector.fg_encoder.layer4.1.bn1.running_mean, %self.kp_detector.fg_encoder.layer4.1.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1/__module.kp_detector.fg_encoder.layer4.1.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %275 : Tensor = aten::relu(%input.217), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1/__module.kp_detector.fg_encoder.layer4.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0.weight, %input.221 : Tensor = aten::_convolution(%275, %self.kp_detector.fg_encoder.layer4.1.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1/__module.kp_detector.fg_encoder.layer4.1.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.31 : Tensor = aten::batch_norm(%input.221, %self.kp_detector.fg_encoder.layer4.1.bn2.weight, %self.kp_detector.fg_encoder.layer4.1.bn2.bias, %self.kp_detector.fg_encoder.layer4.1.bn2.running_mean, %self.kp_detector.fg_encoder.layer4.1.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1/__module.kp_detector.fg_encoder.layer4.1.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %278 : Tensor = aten::add(%out.31, %272, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %279 : Tensor = aten::relu(%278), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1/__module.kp_detector.fg_encoder.layer4.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0.weight, %input.23 : Tensor = aten::_convolution(%55, %self.kp_detector.fg_encoder.layer1.1.conv1.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.1/__module.kp_detector.fg_encoder.layer1.1.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.25 : Tensor = aten::batch_norm(%input.23, %self.kp_detector.fg_encoder.layer1.1.bn1.weight, %self.kp_detector.fg_encoder.layer1.1.bn1.bias, %self.kp_detector.fg_encoder.layer1.1.bn1.running_mean, %self.kp_detector.fg_encoder.layer1.1.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.1/__module.kp_detector.fg_encoder.layer1.1.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %63 : Tensor = aten::relu(%input.25), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.1/__module.kp_detector.fg_encoder.layer1.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0.weight, %input.239 : Tensor = aten::_convolution(%input.237, %self.kp_detector.fg_encoder.layer1.0.conv1.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.0/__module.kp_detector.fg_encoder.layer1.0.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.241 : Tensor = aten::batch_norm(%input.239, %self.kp_detector.fg_encoder.layer1.0.bn1.weight, %self.kp_detector.fg_encoder.layer1.0.bn1.bias, %self.kp_detector.fg_encoder.layer1.0.bn1.running_mean, %self.kp_detector.fg_encoder.layer1.0.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.0/__module.kp_detector.fg_encoder.layer1.0.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %299 : Tensor = aten::relu(%input.241), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.0/__module.kp_detector.fg_encoder.layer1.0.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0.weight, %input.245 : Tensor = aten::_convolution(%299, %self.kp_detector.fg_encoder.layer1.0.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.0/__module.kp_detector.fg_encoder.layer1.0.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.33 : Tensor = aten::batch_norm(%input.245, %self.kp_detector.fg_encoder.layer1.0.bn2.weight, %self.kp_detector.fg_encoder.layer1.0.bn2.bias, %self.kp_detector.fg_encoder.layer1.0.bn2.running_mean, %self.kp_detector.fg_encoder.layer1.0.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.0/__module.kp_detector.fg_encoder.layer1.0.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %302 : Tensor = aten::add(%out.33, %input.237, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.0 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %303 : Tensor = aten::relu(%302), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.0/__module.kp_detector.fg_encoder.layer1.0.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0.weight, %input.251 : Tensor = aten::_convolution(%303, %self.kp_detector.fg_encoder.layer1.1.conv1.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.1/__module.kp_detector.fg_encoder.layer1.1.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.253 : Tensor = aten::batch_norm(%input.251, %self.kp_detector.fg_encoder.layer1.1.bn1.weight, %self.kp_detector.fg_encoder.layer1.1.bn1.bias, %self.kp_detector.fg_encoder.layer1.1.bn1.running_mean, %self.kp_detector.fg_encoder.layer1.1.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.1/__module.kp_detector.fg_encoder.layer1.1.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %306 : Tensor = aten::relu(%input.253), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.1/__module.kp_detector.fg_encoder.layer1.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0.weight, %input.257 : Tensor = aten::_convolution(%306, %self.kp_detector.fg_encoder.layer1.1.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.1/__module.kp_detector.fg_encoder.layer1.1.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.35 : Tensor = aten::batch_norm(%input.257, %self.kp_detector.fg_encoder.layer1.1.bn2.weight, %self.kp_detector.fg_encoder.layer1.1.bn2.bias, %self.kp_detector.fg_encoder.layer1.1.bn2.running_mean, %self.kp_detector.fg_encoder.layer1.1.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.1/__module.kp_detector.fg_encoder.layer1.1.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %309 : Tensor = aten::add(%out.35, %303, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.1 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %310 : Tensor = aten::relu(%309), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.1/__module.kp_detector.fg_encoder.layer1.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0.weight, %input.263 : Tensor = aten::_convolution(%310, %self.kp_detector.fg_encoder.layer2.0.conv1.weight, %5, %25, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.0/__module.kp_detector.fg_encoder.layer2.0.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.265 : Tensor = aten::batch_norm(%input.263, %self.kp_detector.fg_encoder.layer2.0.bn1.weight, %self.kp_detector.fg_encoder.layer2.0.bn1.bias, %self.kp_detector.fg_encoder.layer2.0.bn1.running_mean, %self.kp_detector.fg_encoder.layer2.0.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.0/__module.kp_detector.fg_encoder.layer2.0.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %313 : Tensor = aten::relu(%input.265), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.0/__module.kp_detector.fg_encoder.layer2.0.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0.weight, %input.269 : Tensor = aten::_convolution(%313, %self.kp_detector.fg_encoder.layer2.0.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.0/__module.kp_detector.fg_encoder.layer2.0.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.37 : Tensor = aten::batch_norm(%input.269, %self.kp_detector.fg_encoder.layer2.0.bn2.weight, %self.kp_detector.fg_encoder.layer2.0.bn2.bias, %self.kp_detector.fg_encoder.layer2.0.bn2.running_mean, %self.kp_detector.fg_encoder.layer2.0.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.0/__module.kp_detector.fg_encoder.layer2.0.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0.weight, %input.271 : Tensor = aten::_convolution(%310, %self.kp_detector.fg_encoder.layer2.0.downsample.0.weight, %5, %25, %28, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.0/__module.kp_detector.fg_encoder.layer2.0.downsample/__module.kp_detector.fg_encoder.layer2.0.downsample.0 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %identity.13 : Tensor = aten::batch_norm(%input.271, %self.kp_detector.fg_encoder.layer2.0.downsample.1.weight, %self.kp_detector.fg_encoder.layer2.0.bn2.bias, %self.kp_detector.fg_encoder.layer2.0.downsample.1.running_mean, %self.kp_detector.fg_encoder.layer2.0.downsample.1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.0/__module.kp_detector.fg_encoder.layer2.0.downsample/__module.kp_detector.fg_encoder.layer2.0.downsample.1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %318 : Tensor = aten::add(%out.37, %identity.13, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.0 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %319 : Tensor = aten::relu(%318), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.0/__module.kp_detector.fg_encoder.layer2.0.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0.bias, %input.271 : Tensor = aten::_convolution(%310, %self.kp_detector.fg_encoder.layer2.0.downsample.0.weight, %5, %25, %28, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.0/__module.kp_detector.fg_encoder.layer2.0.downsample/__module.kp_detector.fg_encoder.layer2.0.downsample.0 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %identity.13 : Tensor = aten::batch_norm(%input.271, %self.kp_detector.fg_encoder.layer2.0.downsample.1.weight, %self.kp_detector.fg_encoder.layer2.0.bn2.bias, %self.kp_detector.fg_encoder.layer2.0.downsample.1.running_mean, %self.kp_detector.fg_encoder.layer2.0.downsample.1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.0/__module.kp_detector.fg_encoder.layer2.0.downsample/__module.kp_detector.fg_encoder.layer2.0.downsample.1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %318 : Tensor = aten::add(%out.37, %identity.13, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.0 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %319 : Tensor = aten::relu(%318), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.0/__module.kp_detector.fg_encoder.layer2.0.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0.weight, %input.277 : Tensor = aten::_convolution(%319, %self.kp_detector.fg_encoder.layer2.1.conv1.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.1/__module.kp_detector.fg_encoder.layer2.1.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.279 : Tensor = aten::batch_norm(%input.277, %self.kp_detector.fg_encoder.layer2.1.bn1.weight, %self.kp_detector.fg_encoder.layer2.1.bn1.bias, %self.kp_detector.fg_encoder.layer2.1.bn1.running_mean, %self.kp_detector.fg_encoder.layer2.1.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.1/__module.kp_detector.fg_encoder.layer2.1.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %322 : Tensor = aten::relu(%input.279), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.1/__module.kp_detector.fg_encoder.layer2.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0.weight, %input.283 : Tensor = aten::_convolution(%322, %self.kp_detector.fg_encoder.layer2.1.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.1/__module.kp_detector.fg_encoder.layer2.1.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.39 : Tensor = aten::batch_norm(%input.283, %self.kp_detector.fg_encoder.layer2.1.bn2.weight, %self.kp_detector.fg_encoder.layer2.1.bn2.bias, %self.kp_detector.fg_encoder.layer2.1.bn2.running_mean, %self.kp_detector.fg_encoder.layer2.1.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.1/__module.kp_detector.fg_encoder.layer2.1.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %325 : Tensor = aten::add(%out.39, %319, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.1 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %326 : Tensor = aten::relu(%325), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.1/__module.kp_detector.fg_encoder.layer2.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0.weight, %input.289 : Tensor = aten::_convolution(%326, %self.kp_detector.fg_encoder.layer3.0.conv1.weight, %5, %25, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.291 : Tensor = aten::batch_norm(%input.289, %self.kp_detector.fg_encoder.layer3.0.bn1.weight, %self.kp_detector.fg_encoder.layer3.0.bn1.bias, %self.kp_detector.fg_encoder.layer3.0.bn1.running_mean, %self.kp_detector.fg_encoder.layer3.0.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %329 : Tensor = aten::relu(%input.291), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0.weight, %input.29 : Tensor = aten::_convolution(%63, %self.kp_detector.fg_encoder.layer1.1.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.1/__module.kp_detector.fg_encoder.layer1.1.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.3 : Tensor = aten::batch_norm(%input.29, %self.kp_detector.fg_encoder.layer1.1.bn2.weight, %self.kp_detector.fg_encoder.layer1.1.bn2.bias, %self.kp_detector.fg_encoder.layer1.1.bn2.running_mean, %self.kp_detector.fg_encoder.layer1.1.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.1/__module.kp_detector.fg_encoder.layer1.1.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %71 : Tensor = aten::add(%out.3, %55, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.1 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %72 : Tensor = aten::relu(%71), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.1/__module.kp_detector.fg_encoder.layer1.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0.weight, %input.295 : Tensor = aten::_convolution(%329, %self.kp_detector.fg_encoder.layer3.0.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.41 : Tensor = aten::batch_norm(%input.295, %self.kp_detector.fg_encoder.layer3.0.bn2.weight, %self.kp_detector.fg_encoder.layer3.0.bn2.bias, %self.kp_detector.fg_encoder.layer3.0.bn2.running_mean, %self.kp_detector.fg_encoder.layer3.0.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0.weight, %input.297 : Tensor = aten::_convolution(%326, %self.kp_detector.fg_encoder.layer3.0.downsample.0.weight, %5, %25, %28, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.downsample/__module.kp_detector.fg_encoder.layer3.0.downsample.0 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %identity.15 : Tensor = aten::batch_norm(%input.297, %self.kp_detector.fg_encoder.layer3.0.downsample.1.weight, %self.kp_detector.fg_encoder.layer3.0.bn2.bias, %self.kp_detector.fg_encoder.layer3.0.downsample.1.running_mean, %self.kp_detector.fg_encoder.layer3.0.downsample.1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.downsample/__module.kp_detector.fg_encoder.layer3.0.downsample.1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %334 : Tensor = aten::add(%out.41, %identity.15, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %335 : Tensor = aten::relu(%334), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0.weight, %input.303 : Tensor = aten::_convolution(%335, %self.kp_detector.fg_encoder.layer3.1.conv1.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1/__module.kp_detector.fg_encoder.layer3.1.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.305 : Tensor = aten::batch_norm(%input.303, %self.kp_detector.fg_encoder.layer3.1.bn1.weight, %self.kp_detector.fg_encoder.layer3.1.bn1.bias, %self.kp_detector.fg_encoder.layer3.1.bn1.running_mean, %self.kp_detector.fg_encoder.layer3.1.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1/__module.kp_detector.fg_encoder.layer3.1.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %338 : Tensor = aten::relu(%input.305), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1/__module.kp_detector.fg_encoder.layer3.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0.weight, %input.309 : Tensor = aten::_convolution(%338, %self.kp_detector.fg_encoder.layer3.1.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1/__module.kp_detector.fg_encoder.layer3.1.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.43 : Tensor = aten::batch_norm(%input.309, %self.kp_detector.fg_encoder.layer3.1.bn2.weight, %self.kp_detector.fg_encoder.layer3.1.bn2.bias, %self.kp_detector.fg_encoder.layer3.1.bn2.running_mean, %self.kp_detector.fg_encoder.layer3.1.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1/__module.kp_detector.fg_encoder.layer3.1.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %341 : Tensor = aten::add(%out.43, %335, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %342 : Tensor = aten::relu(%341), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1/__module.kp_detector.fg_encoder.layer3.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0.weight, %input.315 : Tensor = aten::_convolution(%342, %self.kp_detector.fg_encoder.layer4.0.conv1.weight, %5, %25, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.317 : Tensor = aten::batch_norm(%input.315, %self.kp_detector.fg_encoder.layer4.0.bn1.weight, %self.kp_detector.fg_encoder.layer4.0.bn1.bias, %self.kp_detector.fg_encoder.layer4.0.bn1.running_mean, %self.kp_detector.fg_encoder.layer4.0.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %345 : Tensor = aten::relu(%input.317), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0.weight, %input.321 : Tensor = aten::_convolution(%345, %self.kp_detector.fg_encoder.layer4.0.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.45 : Tensor = aten::batch_norm(%input.321, %self.kp_detector.fg_encoder.layer4.0.bn2.weight, %self.kp_detector.fg_encoder.layer4.0.bn2.bias, %self.kp_detector.fg_encoder.layer4.0.bn2.running_mean, %self.kp_detector.fg_encoder.layer4.0.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0.weight, %input.323 : Tensor = aten::_convolution(%342, %self.kp_detector.fg_encoder.layer4.0.downsample.0.weight, %5, %25, %28, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.downsample/__module.kp_detector.fg_encoder.layer4.0.downsample.0 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %identity : Tensor = aten::batch_norm(%input.323, %self.kp_detector.fg_encoder.layer4.0.downsample.1.weight, %self.kp_detector.fg_encoder.layer4.0.bn2.bias, %self.kp_detector.fg_encoder.layer4.0.downsample.1.running_mean, %self.kp_detector.fg_encoder.layer4.0.downsample.1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.downsample/__module.kp_detector.fg_encoder.layer4.0.downsample.1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %350 : Tensor = aten::add(%out.45, %identity, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %351 : Tensor = aten::relu(%350), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0.weight, %input.329 : Tensor = aten::_convolution(%351, %self.kp_detector.fg_encoder.layer4.1.conv1.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1/__module.kp_detector.fg_encoder.layer4.1.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.331 : Tensor = aten::batch_norm(%input.329, %self.kp_detector.fg_encoder.layer4.1.bn1.weight, %self.kp_detector.fg_encoder.layer4.1.bn1.bias, %self.kp_detector.fg_encoder.layer4.1.bn1.running_mean, %self.kp_detector.fg_encoder.layer4.1.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1/__module.kp_detector.fg_encoder.layer4.1.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %354 : Tensor = aten::relu(%input.331), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1/__module.kp_detector.fg_encoder.layer4.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0.weight, %input.335 : Tensor = aten::_convolution(%354, %self.kp_detector.fg_encoder.layer4.1.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1/__module.kp_detector.fg_encoder.layer4.1.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.47 : Tensor = aten::batch_norm(%input.335, %self.kp_detector.fg_encoder.layer4.1.bn2.weight, %self.kp_detector.fg_encoder.layer4.1.bn2.bias, %self.kp_detector.fg_encoder.layer4.1.bn2.running_mean, %self.kp_detector.fg_encoder.layer4.1.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1/__module.kp_detector.fg_encoder.layer4.1.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %357 : Tensor = aten::add(%out.47, %351, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %358 : Tensor = aten::relu(%357), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1/__module.kp_detector.fg_encoder.layer4.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0.weight, %input.35 : Tensor = aten::_convolution(%72, %self.kp_detector.fg_encoder.layer2.0.conv1.weight, %5, %25, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.0/__module.kp_detector.fg_encoder.layer2.0.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.37 : Tensor = aten::batch_norm(%input.35, %self.kp_detector.fg_encoder.layer2.0.bn1.weight, %self.kp_detector.fg_encoder.layer2.0.bn1.bias, %self.kp_detector.fg_encoder.layer2.0.bn1.running_mean, %self.kp_detector.fg_encoder.layer2.0.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.0/__module.kp_detector.fg_encoder.layer2.0.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %80 : Tensor = aten::relu(%input.37), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.0/__module.kp_detector.fg_encoder.layer2.0.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0.weight, %input.41 : Tensor = aten::_convolution(%80, %self.kp_detector.fg_encoder.layer2.0.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.0/__module.kp_detector.fg_encoder.layer2.0.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.5 : Tensor = aten::batch_norm(%input.41, %self.kp_detector.fg_encoder.layer2.0.bn2.weight, %self.kp_detector.fg_encoder.layer2.0.bn2.bias, %self.kp_detector.fg_encoder.layer2.0.bn2.running_mean, %self.kp_detector.fg_encoder.layer2.0.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.0/__module.kp_detector.fg_encoder.layer2.0.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0.weight, %input.43 : Tensor = aten::_convolution(%72, %self.kp_detector.fg_encoder.layer2.0.downsample.0.weight, %5, %25, %28, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.0/__module.kp_detector.fg_encoder.layer2.0.downsample/__module.kp_detector.fg_encoder.layer2.0.downsample.0 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %identity.1 : Tensor = aten::batch_norm(%input.43, %self.kp_detector.fg_encoder.layer2.0.downsample.1.weight, %self.kp_detector.fg_encoder.layer2.0.bn2.bias, %self.kp_detector.fg_encoder.layer2.0.downsample.1.running_mean, %self.kp_detector.fg_encoder.layer2.0.downsample.1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.0/__module.kp_detector.fg_encoder.layer2.0.downsample/__module.kp_detector.fg_encoder.layer2.0.downsample.1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %94 : Tensor = aten::add(%out.5, %identity.1, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.0 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %95 : Tensor = aten::relu(%94), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.0/__module.kp_detector.fg_encoder.layer2.0.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0.bias, %input.43 : Tensor = aten::_convolution(%72, %self.kp_detector.fg_encoder.layer2.0.downsample.0.weight, %5, %25, %28, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.0/__module.kp_detector.fg_encoder.layer2.0.downsample/__module.kp_detector.fg_encoder.layer2.0.downsample.0 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %identity.1 : Tensor = aten::batch_norm(%input.43, %self.kp_detector.fg_encoder.layer2.0.downsample.1.weight, %self.kp_detector.fg_encoder.layer2.0.bn2.bias, %self.kp_detector.fg_encoder.layer2.0.downsample.1.running_mean, %self.kp_detector.fg_encoder.layer2.0.downsample.1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.0/__module.kp_detector.fg_encoder.layer2.0.downsample/__module.kp_detector.fg_encoder.layer2.0.downsample.1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %94 : Tensor = aten::add(%out.5, %identity.1, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.0 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %95 : Tensor = aten::relu(%94), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.0/__module.kp_detector.fg_encoder.layer2.0.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0.weight, %input.49 : Tensor = aten::_convolution(%95, %self.kp_detector.fg_encoder.layer2.1.conv1.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.1/__module.kp_detector.fg_encoder.layer2.1.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.51 : Tensor = aten::batch_norm(%input.49, %self.kp_detector.fg_encoder.layer2.1.bn1.weight, %self.kp_detector.fg_encoder.layer2.1.bn1.bias, %self.kp_detector.fg_encoder.layer2.1.bn1.running_mean, %self.kp_detector.fg_encoder.layer2.1.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.1/__module.kp_detector.fg_encoder.layer2.1.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %103 : Tensor = aten::relu(%input.51), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.1/__module.kp_detector.fg_encoder.layer2.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0.weight, %input.55 : Tensor = aten::_convolution(%103, %self.kp_detector.fg_encoder.layer2.1.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.1/__module.kp_detector.fg_encoder.layer2.1.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.7 : Tensor = aten::batch_norm(%input.55, %self.kp_detector.fg_encoder.layer2.1.bn2.weight, %self.kp_detector.fg_encoder.layer2.1.bn2.bias, %self.kp_detector.fg_encoder.layer2.1.bn2.running_mean, %self.kp_detector.fg_encoder.layer2.1.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.1/__module.kp_detector.fg_encoder.layer2.1.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %111 : Tensor = aten::add(%out.7, %95, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.1 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %112 : Tensor = aten::relu(%111), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.1/__module.kp_detector.fg_encoder.layer2.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0.weight, %input.61 : Tensor = aten::_convolution(%112, %self.kp_detector.fg_encoder.layer3.0.conv1.weight, %5, %25, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.63 : Tensor = aten::batch_norm(%input.61, %self.kp_detector.fg_encoder.layer3.0.bn1.weight, %self.kp_detector.fg_encoder.layer3.0.bn1.bias, %self.kp_detector.fg_encoder.layer3.0.bn1.running_mean, %self.kp_detector.fg_encoder.layer3.0.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %120 : Tensor = aten::relu(%input.63), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0.weight, %input.67 : Tensor = aten::_convolution(%120, %self.kp_detector.fg_encoder.layer3.0.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.9 : Tensor = aten::batch_norm(%input.67, %self.kp_detector.fg_encoder.layer3.0.bn2.weight, %self.kp_detector.fg_encoder.layer3.0.bn2.bias, %self.kp_detector.fg_encoder.layer3.0.bn2.running_mean, %self.kp_detector.fg_encoder.layer3.0.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0.weight, %input.69 : Tensor = aten::_convolution(%112, %self.kp_detector.fg_encoder.layer3.0.downsample.0.weight, %5, %25, %28, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.downsample/__module.kp_detector.fg_encoder.layer3.0.downsample.0 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %identity.3 : Tensor = aten::batch_norm(%input.69, %self.kp_detector.fg_encoder.layer3.0.downsample.1.weight, %self.kp_detector.fg_encoder.layer3.0.bn2.bias, %self.kp_detector.fg_encoder.layer3.0.downsample.1.running_mean, %self.kp_detector.fg_encoder.layer3.0.downsample.1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.downsample/__module.kp_detector.fg_encoder.layer3.0.downsample.1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %134 : Tensor = aten::add(%out.9, %identity.3, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %135 : Tensor = aten::relu(%134), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0.weight, %input.75 : Tensor = aten::_convolution(%135, %self.kp_detector.fg_encoder.layer3.1.conv1.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1/__module.kp_detector.fg_encoder.layer3.1.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.77 : Tensor = aten::batch_norm(%input.75, %self.kp_detector.fg_encoder.layer3.1.bn1.weight, %self.kp_detector.fg_encoder.layer3.1.bn1.bias, %self.kp_detector.fg_encoder.layer3.1.bn1.running_mean, %self.kp_detector.fg_encoder.layer3.1.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1/__module.kp_detector.fg_encoder.layer3.1.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %143 : Tensor = aten::relu(%input.77), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1/__module.kp_detector.fg_encoder.layer3.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0.weight, %input.81 : Tensor = aten::_convolution(%143, %self.kp_detector.fg_encoder.layer3.1.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1/__module.kp_detector.fg_encoder.layer3.1.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.11 : Tensor = aten::batch_norm(%input.81, %self.kp_detector.fg_encoder.layer3.1.bn2.weight, %self.kp_detector.fg_encoder.layer3.1.bn2.bias, %self.kp_detector.fg_encoder.layer3.1.bn2.running_mean, %self.kp_detector.fg_encoder.layer3.1.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1/__module.kp_detector.fg_encoder.layer3.1.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %151 : Tensor = aten::add(%out.11, %135, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %152 : Tensor = aten::relu(%151), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1/__module.kp_detector.fg_encoder.layer3.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0.weight, %input.87 : Tensor = aten::_convolution(%152, %self.kp_detector.fg_encoder.layer4.0.conv1.weight, %5, %25, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.89 : Tensor = aten::batch_norm(%input.87, %self.kp_detector.fg_encoder.layer4.0.bn1.weight, %self.kp_detector.fg_encoder.layer4.0.bn1.bias, %self.kp_detector.fg_encoder.layer4.0.bn1.running_mean, %self.kp_detector.fg_encoder.layer4.0.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %160 : Tensor = aten::relu(%input.89), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0.weight, %input.93 : Tensor = aten::_convolution(%160, %self.kp_detector.fg_encoder.layer4.0.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.13 : Tensor = aten::batch_norm(%input.93, %self.kp_detector.fg_encoder.layer4.0.bn2.weight, %self.kp_detector.fg_encoder.layer4.0.bn2.bias, %self.kp_detector.fg_encoder.layer4.0.bn2.running_mean, %self.kp_detector.fg_encoder.layer4.0.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0.weight, %input.95 : Tensor = aten::_convolution(%152, %self.kp_detector.fg_encoder.layer4.0.downsample.0.weight, %5, %25, %28, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.downsample/__module.kp_detector.fg_encoder.layer4.0.downsample.0 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %identity.5 : Tensor = aten::batch_norm(%input.95, %self.kp_detector.fg_encoder.layer4.0.downsample.1.weight, %self.kp_detector.fg_encoder.layer4.0.bn2.bias, %self.kp_detector.fg_encoder.layer4.0.downsample.1.running_mean, %self.kp_detector.fg_encoder.layer4.0.downsample.1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.downsample/__module.kp_detector.fg_encoder.layer4.0.downsample.1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %174 : Tensor = aten::add(%out.13, %identity.5, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %175 : Tensor = aten::relu(%174), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0.weight
WARNING: [Torch-TensorRT TorchScript Conversion Context] - - 27 weights are affected by this issue: Detected values less than smallest positive FP16 subnormal value and converted them to the FP16 minimum subnormalized value.
DEBUG: [Torch-TensorRT TorchScript Conversion Context] -   List of affected weights: %input.101 : Tensor = aten::_convolution(%175, %self.kp_detector.fg_encoder.layer4.1.conv1.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1/__module.kp_detector.fg_encoder.layer4.1.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.103 : Tensor = aten::batch_norm(%input.101, %self.kp_detector.fg_encoder.layer4.1.bn1.weight, %self.kp_detector.fg_encoder.layer4.1.bn1.bias, %self.kp_detector.fg_encoder.layer4.1.bn1.running_mean, %self.kp_detector.fg_encoder.layer4.1.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1/__module.kp_detector.fg_encoder.layer4.1.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %183 : Tensor = aten::relu(%input.103), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1/__module.kp_detector.fg_encoder.layer4.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0.bias, %input.101 : Tensor = aten::_convolution(%175, %self.kp_detector.fg_encoder.layer4.1.conv1.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1/__module.kp_detector.fg_encoder.layer4.1.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.103 : Tensor = aten::batch_norm(%input.101, %self.kp_detector.fg_encoder.layer4.1.bn1.weight, %self.kp_detector.fg_encoder.layer4.1.bn1.bias, %self.kp_detector.fg_encoder.layer4.1.bn1.running_mean, %self.kp_detector.fg_encoder.layer4.1.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1/__module.kp_detector.fg_encoder.layer4.1.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %183 : Tensor = aten::relu(%input.103), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1/__module.kp_detector.fg_encoder.layer4.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0.weight, %input.107 : Tensor = aten::_convolution(%183, %self.kp_detector.fg_encoder.layer4.1.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1/__module.kp_detector.fg_encoder.layer4.1.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.15 : Tensor = aten::batch_norm(%input.107, %self.kp_detector.fg_encoder.layer4.1.bn2.weight, %self.kp_detector.fg_encoder.layer4.1.bn2.bias, %self.kp_detector.fg_encoder.layer4.1.bn2.running_mean, %self.kp_detector.fg_encoder.layer4.1.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1/__module.kp_detector.fg_encoder.layer4.1.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %191 : Tensor = aten::add(%out.15, %175, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %192 : Tensor = aten::relu(%191), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1/__module.kp_detector.fg_encoder.layer4.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0.weight, %input.131 : Tensor = aten::_convolution(%220, %self.kp_detector.fg_encoder.layer1.0.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.0/__module.kp_detector.fg_encoder.layer1.0.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.17 : Tensor = aten::batch_norm(%input.131, %self.kp_detector.fg_encoder.layer1.0.bn2.weight, %self.kp_detector.fg_encoder.layer1.0.bn2.bias, %self.kp_detector.fg_encoder.layer1.0.bn2.running_mean, %self.kp_detector.fg_encoder.layer1.0.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.0/__module.kp_detector.fg_encoder.layer1.0.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %223 : Tensor = aten::add(%out.17, %input.123, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.0 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %224 : Tensor = aten::relu(%223), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.0/__module.kp_detector.fg_encoder.layer1.0.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0.weight, %input.17 : Tensor = aten::_convolution(%46, %self.kp_detector.fg_encoder.layer1.0.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.0/__module.kp_detector.fg_encoder.layer1.0.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.1 : Tensor = aten::batch_norm(%input.17, %self.kp_detector.fg_encoder.layer1.0.bn2.weight, %self.kp_detector.fg_encoder.layer1.0.bn2.bias, %self.kp_detector.fg_encoder.layer1.0.bn2.running_mean, %self.kp_detector.fg_encoder.layer1.0.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.0/__module.kp_detector.fg_encoder.layer1.0.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %54 : Tensor = aten::add(%out.1, %input.9, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.0 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %55 : Tensor = aten::relu(%54), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.0/__module.kp_detector.fg_encoder.layer1.0.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0.weight, %input.181 : Tensor = aten::_convolution(%250, %self.kp_detector.fg_encoder.layer3.0.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.25 : Tensor = aten::batch_norm(%input.181, %self.kp_detector.fg_encoder.layer3.0.bn2.weight, %self.kp_detector.fg_encoder.layer3.0.bn2.bias, %self.kp_detector.fg_encoder.layer3.0.bn2.running_mean, %self.kp_detector.fg_encoder.layer3.0.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0.weight, %input.189 : Tensor = aten::_convolution(%256, %self.kp_detector.fg_encoder.layer3.1.conv1.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1/__module.kp_detector.fg_encoder.layer3.1.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.191 : Tensor = aten::batch_norm(%input.189, %self.kp_detector.fg_encoder.layer3.1.bn1.weight, %self.kp_detector.fg_encoder.layer3.1.bn1.bias, %self.kp_detector.fg_encoder.layer3.1.bn1.running_mean, %self.kp_detector.fg_encoder.layer3.1.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1/__module.kp_detector.fg_encoder.layer3.1.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %259 : Tensor = aten::relu(%input.191), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1/__module.kp_detector.fg_encoder.layer3.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0.weight, %input.195 : Tensor = aten::_convolution(%259, %self.kp_detector.fg_encoder.layer3.1.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1/__module.kp_detector.fg_encoder.layer3.1.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.27 : Tensor = aten::batch_norm(%input.195, %self.kp_detector.fg_encoder.layer3.1.bn2.weight, %self.kp_detector.fg_encoder.layer3.1.bn2.bias, %self.kp_detector.fg_encoder.layer3.1.bn2.running_mean, %self.kp_detector.fg_encoder.layer3.1.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1/__module.kp_detector.fg_encoder.layer3.1.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %262 : Tensor = aten::add(%out.27, %256, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %263 : Tensor = aten::relu(%262), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1/__module.kp_detector.fg_encoder.layer3.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0.weight, %input.201 : Tensor = aten::_convolution(%263, %self.kp_detector.fg_encoder.layer4.0.conv1.weight, %5, %25, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.203 : Tensor = aten::batch_norm(%input.201, %self.kp_detector.fg_encoder.layer4.0.bn1.weight, %self.kp_detector.fg_encoder.layer4.0.bn1.bias, %self.kp_detector.fg_encoder.layer4.0.bn1.running_mean, %self.kp_detector.fg_encoder.layer4.0.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %266 : Tensor = aten::relu(%input.203), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0.weight, %input.207 : Tensor = aten::_convolution(%266, %self.kp_detector.fg_encoder.layer4.0.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.29 : Tensor = aten::batch_norm(%input.207, %self.kp_detector.fg_encoder.layer4.0.bn2.weight, %self.kp_detector.fg_encoder.layer4.0.bn2.bias, %self.kp_detector.fg_encoder.layer4.0.bn2.running_mean, %self.kp_detector.fg_encoder.layer4.0.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0.weight, %input.215 : Tensor = aten::_convolution(%272, %self.kp_detector.fg_encoder.layer4.1.conv1.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1/__module.kp_detector.fg_encoder.layer4.1.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.217 : Tensor = aten::batch_norm(%input.215, %self.kp_detector.fg_encoder.layer4.1.bn1.weight, %self.kp_detector.fg_encoder.layer4.1.bn1.bias, %self.kp_detector.fg_encoder.layer4.1.bn1.running_mean, %self.kp_detector.fg_encoder.layer4.1.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1/__module.kp_detector.fg_encoder.layer4.1.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %275 : Tensor = aten::relu(%input.217), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1/__module.kp_detector.fg_encoder.layer4.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0.bias, %input.215 : Tensor = aten::_convolution(%272, %self.kp_detector.fg_encoder.layer4.1.conv1.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1/__module.kp_detector.fg_encoder.layer4.1.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.217 : Tensor = aten::batch_norm(%input.215, %self.kp_detector.fg_encoder.layer4.1.bn1.weight, %self.kp_detector.fg_encoder.layer4.1.bn1.bias, %self.kp_detector.fg_encoder.layer4.1.bn1.running_mean, %self.kp_detector.fg_encoder.layer4.1.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1/__module.kp_detector.fg_encoder.layer4.1.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %275 : Tensor = aten::relu(%input.217), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1/__module.kp_detector.fg_encoder.layer4.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0.weight, %input.221 : Tensor = aten::_convolution(%275, %self.kp_detector.fg_encoder.layer4.1.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1/__module.kp_detector.fg_encoder.layer4.1.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.31 : Tensor = aten::batch_norm(%input.221, %self.kp_detector.fg_encoder.layer4.1.bn2.weight, %self.kp_detector.fg_encoder.layer4.1.bn2.bias, %self.kp_detector.fg_encoder.layer4.1.bn2.running_mean, %self.kp_detector.fg_encoder.layer4.1.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1/__module.kp_detector.fg_encoder.layer4.1.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %278 : Tensor = aten::add(%out.31, %272, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %279 : Tensor = aten::relu(%278), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1/__module.kp_detector.fg_encoder.layer4.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0.weight, %input.245 : Tensor = aten::_convolution(%299, %self.kp_detector.fg_encoder.layer1.0.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.0/__module.kp_detector.fg_encoder.layer1.0.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.33 : Tensor = aten::batch_norm(%input.245, %self.kp_detector.fg_encoder.layer1.0.bn2.weight, %self.kp_detector.fg_encoder.layer1.0.bn2.bias, %self.kp_detector.fg_encoder.layer1.0.bn2.running_mean, %self.kp_detector.fg_encoder.layer1.0.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.0/__module.kp_detector.fg_encoder.layer1.0.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %302 : Tensor = aten::add(%out.33, %input.237, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.0 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %303 : Tensor = aten::relu(%302), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.0/__module.kp_detector.fg_encoder.layer1.0.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0.weight, %input.295 : Tensor = aten::_convolution(%329, %self.kp_detector.fg_encoder.layer3.0.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.41 : Tensor = aten::batch_norm(%input.295, %self.kp_detector.fg_encoder.layer3.0.bn2.weight, %self.kp_detector.fg_encoder.layer3.0.bn2.bias, %self.kp_detector.fg_encoder.layer3.0.bn2.running_mean, %self.kp_detector.fg_encoder.layer3.0.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0.weight, %input.303 : Tensor = aten::_convolution(%335, %self.kp_detector.fg_encoder.layer3.1.conv1.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1/__module.kp_detector.fg_encoder.layer3.1.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.305 : Tensor = aten::batch_norm(%input.303, %self.kp_detector.fg_encoder.layer3.1.bn1.weight, %self.kp_detector.fg_encoder.layer3.1.bn1.bias, %self.kp_detector.fg_encoder.layer3.1.bn1.running_mean, %self.kp_detector.fg_encoder.layer3.1.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1/__module.kp_detector.fg_encoder.layer3.1.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %338 : Tensor = aten::relu(%input.305), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1/__module.kp_detector.fg_encoder.layer3.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0.weight, %input.309 : Tensor = aten::_convolution(%338, %self.kp_detector.fg_encoder.layer3.1.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1/__module.kp_detector.fg_encoder.layer3.1.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.43 : Tensor = aten::batch_norm(%input.309, %self.kp_detector.fg_encoder.layer3.1.bn2.weight, %self.kp_detector.fg_encoder.layer3.1.bn2.bias, %self.kp_detector.fg_encoder.layer3.1.bn2.running_mean, %self.kp_detector.fg_encoder.layer3.1.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1/__module.kp_detector.fg_encoder.layer3.1.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %341 : Tensor = aten::add(%out.43, %335, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %342 : Tensor = aten::relu(%341), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1/__module.kp_detector.fg_encoder.layer3.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0.weight, %input.315 : Tensor = aten::_convolution(%342, %self.kp_detector.fg_encoder.layer4.0.conv1.weight, %5, %25, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.317 : Tensor = aten::batch_norm(%input.315, %self.kp_detector.fg_encoder.layer4.0.bn1.weight, %self.kp_detector.fg_encoder.layer4.0.bn1.bias, %self.kp_detector.fg_encoder.layer4.0.bn1.running_mean, %self.kp_detector.fg_encoder.layer4.0.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %345 : Tensor = aten::relu(%input.317), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0.weight, %input.321 : Tensor = aten::_convolution(%345, %self.kp_detector.fg_encoder.layer4.0.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.45 : Tensor = aten::batch_norm(%input.321, %self.kp_detector.fg_encoder.layer4.0.bn2.weight, %self.kp_detector.fg_encoder.layer4.0.bn2.bias, %self.kp_detector.fg_encoder.layer4.0.bn2.running_mean, %self.kp_detector.fg_encoder.layer4.0.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0.weight, %input.329 : Tensor = aten::_convolution(%351, %self.kp_detector.fg_encoder.layer4.1.conv1.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1/__module.kp_detector.fg_encoder.layer4.1.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.331 : Tensor = aten::batch_norm(%input.329, %self.kp_detector.fg_encoder.layer4.1.bn1.weight, %self.kp_detector.fg_encoder.layer4.1.bn1.bias, %self.kp_detector.fg_encoder.layer4.1.bn1.running_mean, %self.kp_detector.fg_encoder.layer4.1.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1/__module.kp_detector.fg_encoder.layer4.1.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %354 : Tensor = aten::relu(%input.331), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1/__module.kp_detector.fg_encoder.layer4.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0.bias, %input.329 : Tensor = aten::_convolution(%351, %self.kp_detector.fg_encoder.layer4.1.conv1.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1/__module.kp_detector.fg_encoder.layer4.1.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.331 : Tensor = aten::batch_norm(%input.329, %self.kp_detector.fg_encoder.layer4.1.bn1.weight, %self.kp_detector.fg_encoder.layer4.1.bn1.bias, %self.kp_detector.fg_encoder.layer4.1.bn1.running_mean, %self.kp_detector.fg_encoder.layer4.1.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1/__module.kp_detector.fg_encoder.layer4.1.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %354 : Tensor = aten::relu(%input.331), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1/__module.kp_detector.fg_encoder.layer4.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0.weight, %input.335 : Tensor = aten::_convolution(%354, %self.kp_detector.fg_encoder.layer4.1.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1/__module.kp_detector.fg_encoder.layer4.1.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.47 : Tensor = aten::batch_norm(%input.335, %self.kp_detector.fg_encoder.layer4.1.bn2.weight, %self.kp_detector.fg_encoder.layer4.1.bn2.bias, %self.kp_detector.fg_encoder.layer4.1.bn2.running_mean, %self.kp_detector.fg_encoder.layer4.1.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1/__module.kp_detector.fg_encoder.layer4.1.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %357 : Tensor = aten::add(%out.47, %351, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %358 : Tensor = aten::relu(%357), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1/__module.kp_detector.fg_encoder.layer4.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0.weight, %input.67 : Tensor = aten::_convolution(%120, %self.kp_detector.fg_encoder.layer3.0.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.9 : Tensor = aten::batch_norm(%input.67, %self.kp_detector.fg_encoder.layer3.0.bn2.weight, %self.kp_detector.fg_encoder.layer3.0.bn2.bias, %self.kp_detector.fg_encoder.layer3.0.bn2.running_mean, %self.kp_detector.fg_encoder.layer3.0.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0.weight, %input.75 : Tensor = aten::_convolution(%135, %self.kp_detector.fg_encoder.layer3.1.conv1.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1/__module.kp_detector.fg_encoder.layer3.1.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.77 : Tensor = aten::batch_norm(%input.75, %self.kp_detector.fg_encoder.layer3.1.bn1.weight, %self.kp_detector.fg_encoder.layer3.1.bn1.bias, %self.kp_detector.fg_encoder.layer3.1.bn1.running_mean, %self.kp_detector.fg_encoder.layer3.1.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1/__module.kp_detector.fg_encoder.layer3.1.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %143 : Tensor = aten::relu(%input.77), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1/__module.kp_detector.fg_encoder.layer3.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0.weight, %input.81 : Tensor = aten::_convolution(%143, %self.kp_detector.fg_encoder.layer3.1.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1/__module.kp_detector.fg_encoder.layer3.1.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.11 : Tensor = aten::batch_norm(%input.81, %self.kp_detector.fg_encoder.layer3.1.bn2.weight, %self.kp_detector.fg_encoder.layer3.1.bn2.bias, %self.kp_detector.fg_encoder.layer3.1.bn2.running_mean, %self.kp_detector.fg_encoder.layer3.1.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1/__module.kp_detector.fg_encoder.layer3.1.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %151 : Tensor = aten::add(%out.11, %135, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %152 : Tensor = aten::relu(%151), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1/__module.kp_detector.fg_encoder.layer3.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0.weight, %input.87 : Tensor = aten::_convolution(%152, %self.kp_detector.fg_encoder.layer4.0.conv1.weight, %5, %25, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.89 : Tensor = aten::batch_norm(%input.87, %self.kp_detector.fg_encoder.layer4.0.bn1.weight, %self.kp_detector.fg_encoder.layer4.0.bn1.bias, %self.kp_detector.fg_encoder.layer4.0.bn1.running_mean, %self.kp_detector.fg_encoder.layer4.0.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %160 : Tensor = aten::relu(%input.89), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0.weight, %input.93 : Tensor = aten::_convolution(%160, %self.kp_detector.fg_encoder.layer4.0.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.13 : Tensor = aten::batch_norm(%input.93, %self.kp_detector.fg_encoder.layer4.0.bn2.weight, %self.kp_detector.fg_encoder.layer4.0.bn2.bias, %self.kp_detector.fg_encoder.layer4.0.bn2.running_mean, %self.kp_detector.fg_encoder.layer4.0.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0.weight
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Deleting timing cache: 961 entries, served 4916 hits since creation.
DEBUG: [Torch-TensorRT TorchScript Conversion Context] - Engine Layer Information:
Layer(Myelin): {ForeignNode[(Unnamed Layer* 1) [Shuffle]...%input.115 : Tensor = aten::select(%19, %21, %16) # /home/joy/remote/stable-diffusion-backend/personalize/thin_plate_spline_motion_model/tpsnn.py:529:0]}, Tactic: 0x0000000000000000, input_0 (Float[1,256,256,3]) -> (Unnamed Layer* 273) [Identity]_output (Float[]), (Unnamed Layer* 269) [Shuffle]_output (Float[1,1]), (Unnamed Layer* 261) [Shuffle]_output (Float[1,1]), (Unnamed Layer* 257) [Shuffle]_output (Float[1,1]), (Unnamed Layer* 177) [Shuffle]_output (Float[1,1]), (Unnamed Layer* 173) [Shuffle]_output (Float[1,1]), (Unnamed Layer* 92) [Shuffle]_output (Float[1,1]), (Unnamed Layer* 88) [Shuffle]_output (Float[1,1]), (Unnamed Layer* 10) [Shuffle]_output (Float[1,3,256,256])
Layer(NoOp): unsqueeze_node_after_{ForeignNode[(Unnamed Layer* 1) [Shuffle]...%input.115 : Tensor = aten::select(%19, %21, %16) # /home/joy/remote/stable-diffusion-backend/personalize/thin_plate_spline_motion_model/tpsnn.py:529:0]}_(Unnamed Layer* 173) [Shuffle]_output, Tactic: 0x0000000000000000, (Unnamed Layer* 173) [Shuffle]_output (Float[1,1]) -> unsqueeze_tensor_after_{ForeignNode[(Unnamed Layer* 1) [Shuffle]...%input.115 : Tensor = aten::select(%19, %21, %16) # /home/joy/remote/stable-diffusion-backend/personalize/thin_plate_spline_motion_model/tpsnn.py:529:0]}_(Unnamed Layer* 173) [Shuffle]_output_out_tensor (Float[1,1,1,1])
Layer(NoOp): unsqueeze_node_after_{ForeignNode[(Unnamed Layer* 1) [Shuffle]...%input.115 : Tensor = aten::select(%19, %21, %16) # /home/joy/remote/stable-diffusion-backend/personalize/thin_plate_spline_motion_model/tpsnn.py:529:0]}_(Unnamed Layer* 257) [Shuffle]_output, Tactic: 0x0000000000000000, (Unnamed Layer* 257) [Shuffle]_output (Float[1,1]) -> unsqueeze_tensor_after_{ForeignNode[(Unnamed Layer* 1) [Shuffle]...%input.115 : Tensor = aten::select(%19, %21, %16) # /home/joy/remote/stable-diffusion-backend/personalize/thin_plate_spline_motion_model/tpsnn.py:529:0]}_(Unnamed Layer* 257) [Shuffle]_output_out_tensor (Float[1,1,1,1])
Layer(Reformat): Reformatting CopyNode for Input Tensor 0 to unsqueeze_node_after_{ForeignNode[(Unnamed Layer* 1) [Shuffle]...%input.115 : Tensor = aten::select(%19, %21, %16) # /home/joy/remote/stable-diffusion-backend/personalize/thin_plate_spline_motion_model/tpsnn.py:529:0]}_(Unnamed Layer* 88) [Shuffle]_output, Tactic: 0x00000000000003e8, (Unnamed Layer* 88) [Shuffle]_output (Float[1,1]) -> Reformatted Input Tensor 0 to unsqueeze_node_after_{ForeignNode[(Unnamed Layer* 1) [Shuffle]...%input.115 : Tensor = aten::select(%19, %21, %16) # /home/joy/remote/stable-diffusion-backend/personalize/thin_plate_spline_motion_model/tpsnn.py:529:0]}_(Unnamed Layer* 88) [Shuffle]_output (Half[1,1])
Layer(NoOp): unsqueeze_node_after_{ForeignNode[(Unnamed Layer* 1) [Shuffle]...%input.115 : Tensor = aten::select(%19, %21, %16) # /home/joy/remote/stable-diffusion-backend/personalize/thin_plate_spline_motion_model/tpsnn.py:529:0]}_(Unnamed Layer* 88) [Shuffle]_output, Tactic: 0x0000000000000000, Reformatted Input Tensor 0 to unsqueeze_node_after_{ForeignNode[(Unnamed Layer* 1) [Shuffle]...%input.115 : Tensor = aten::select(%19, %21, %16) # /home/joy/remote/stable-diffusion-backend/personalize/thin_plate_spline_motion_model/tpsnn.py:529:0]}_(Unnamed Layer* 88) [Shuffle]_output (Half[1,1]) -> unsqueeze_tensor_after_{ForeignNode[(Unnamed Layer* 1) [Shuffle]...%input.115 : Tensor = aten::select(%19, %21, %16) # /home/joy/remote/stable-diffusion-backend/personalize/thin_plate_spline_motion_model/tpsnn.py:529:0]}_(Unnamed Layer* 88) [Shuffle]_output_out_tensor (Half[1,1,1,1])
Layer(Shuffle): %source : Tensor = aten::permute(%6, %9) # /home/joy/remote/stable-diffusion-backend/personalize/thin_plate_spline_motion_model/tpsnn.py:524:0, Tactic: 0x0000000000000000, input_1 (Float[1,256,256,3]) -> output_1 (Float[1,3,256,256])
Layer(Reformat): Reformatting CopyNode for Input Tensor 0 to %input.3 : Tensor = aten::_convolution(%10, %self.kp_detector.fg_encoder.conv1.weight, %5, %25, %26, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.5 : Tensor = aten::batch_norm(%input.3, %self.kp_detector.fg_encoder.bn1.weight, %self.kp_detector.fg_encoder.bn1.bias, %self.kp_detector.fg_encoder.bn1.running_mean, %self.kp_detector.fg_encoder.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %37 : Tensor = aten::relu(%input.5), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0, Tactic: 0x0000000000000000, output_1 (Float[1,3,256,256]) -> Reformatted Input Tensor 0 to %input.3 : Tensor = aten::_convolution(%10, %self.kp_detector.fg_encoder.conv1.weight, %5, %25, %26, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.5 : Tensor = aten::batch_norm(%input.3, %self.kp_detector.fg_encoder.bn1.weight, %self.kp_detector.fg_encoder.bn1.bias, %self.kp_detector.fg_encoder.bn1.running_mean, %self.kp_detector.fg_encoder.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %37 : Tensor = aten::relu(%input.5), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 (Half[1,3:8,256,256])
Layer(CaskConvolution): %input.3 : Tensor = aten::_convolution(%10, %self.kp_detector.fg_encoder.conv1.weight, %5, %25, %26, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.5 : Tensor = aten::batch_norm(%input.3, %self.kp_detector.fg_encoder.bn1.weight, %self.kp_detector.fg_encoder.bn1.bias, %self.kp_detector.fg_encoder.bn1.running_mean, %self.kp_detector.fg_encoder.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %37 : Tensor = aten::relu(%input.5), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0, Tactic: 0x603af898ad7e47f4, Reformatted Input Tensor 0 to %input.3 : Tensor = aten::_convolution(%10, %self.kp_detector.fg_encoder.conv1.weight, %5, %25, %26, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.5 : Tensor = aten::batch_norm(%input.3, %self.kp_detector.fg_encoder.bn1.weight, %self.kp_detector.fg_encoder.bn1.bias, %self.kp_detector.fg_encoder.bn1.running_mean, %self.kp_detector.fg_encoder.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %37 : Tensor = aten::relu(%input.5), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 (Half[1,3:8,256,256]) -> (Unnamed Layer* 13) [Activation]_output (Half[1,64:8,128,128])
Layer(CaskPooling): %input.9 : Tensor = aten::max_pool2d(%37, %26, %25, %27, %27, %4), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.maxpool # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:780:0, Tactic: 0xdb415cba6b0e9137, (Unnamed Layer* 13) [Activation]_output (Half[1,64:8,128,128]) -> (Unnamed Layer* 14) [Pooling]_output (Half[1,64:8,64,64])
Layer(CaskConvolution): %input.11 : Tensor = aten::_convolution(%input.9, %self.kp_detector.fg_encoder.layer1.0.conv1.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.0/__module.kp_detector.fg_encoder.layer1.0.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.13 : Tensor = aten::batch_norm(%input.11, %self.kp_detector.fg_encoder.layer1.0.bn1.weight, %self.kp_detector.fg_encoder.layer1.0.bn1.bias, %self.kp_detector.fg_encoder.layer1.0.bn1.running_mean, %self.kp_detector.fg_encoder.layer1.0.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.0/__module.kp_detector.fg_encoder.layer1.0.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %46 : Tensor = aten::relu(%input.13), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.0/__module.kp_detector.fg_encoder.layer1.0.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0, Tactic: 0x30e8a8d7a953e5e9, (Unnamed Layer* 14) [Pooling]_output (Half[1,64:8,64,64]) -> (Unnamed Layer* 17) [Activation]_output (Half[1,64:8,64,64])
Layer(CaskConvolution): %input.17 : Tensor = aten::_convolution(%46, %self.kp_detector.fg_encoder.layer1.0.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.0/__module.kp_detector.fg_encoder.layer1.0.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.1 : Tensor = aten::batch_norm(%input.17, %self.kp_detector.fg_encoder.layer1.0.bn2.weight, %self.kp_detector.fg_encoder.layer1.0.bn2.bias, %self.kp_detector.fg_encoder.layer1.0.bn2.running_mean, %self.kp_detector.fg_encoder.layer1.0.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.0/__module.kp_detector.fg_encoder.layer1.0.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %54 : Tensor = aten::add(%out.1, %input.9, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.0 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %55 : Tensor = aten::relu(%54), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.0/__module.kp_detector.fg_encoder.layer1.0.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0, Tactic: 0x30e8a8d7a953e5e9, (Unnamed Layer* 17) [Activation]_output (Half[1,64:8,64,64]), (Unnamed Layer* 14) [Pooling]_output (Half[1,64:8,64,64]) -> (Unnamed Layer* 21) [Activation]_output (Half[1,64:8,64,64])
Layer(CaskConvolution): %input.23 : Tensor = aten::_convolution(%55, %self.kp_detector.fg_encoder.layer1.1.conv1.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.1/__module.kp_detector.fg_encoder.layer1.1.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.25 : Tensor = aten::batch_norm(%input.23, %self.kp_detector.fg_encoder.layer1.1.bn1.weight, %self.kp_detector.fg_encoder.layer1.1.bn1.bias, %self.kp_detector.fg_encoder.layer1.1.bn1.running_mean, %self.kp_detector.fg_encoder.layer1.1.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.1/__module.kp_detector.fg_encoder.layer1.1.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %63 : Tensor = aten::relu(%input.25), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.1/__module.kp_detector.fg_encoder.layer1.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0, Tactic: 0x30e8a8d7a953e5e9, (Unnamed Layer* 21) [Activation]_output (Half[1,64:8,64,64]) -> (Unnamed Layer* 24) [Activation]_output (Half[1,64:8,64,64])
Layer(CaskConvolution): %input.29 : Tensor = aten::_convolution(%63, %self.kp_detector.fg_encoder.layer1.1.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.1/__module.kp_detector.fg_encoder.layer1.1.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.3 : Tensor = aten::batch_norm(%input.29, %self.kp_detector.fg_encoder.layer1.1.bn2.weight, %self.kp_detector.fg_encoder.layer1.1.bn2.bias, %self.kp_detector.fg_encoder.layer1.1.bn2.running_mean, %self.kp_detector.fg_encoder.layer1.1.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.1/__module.kp_detector.fg_encoder.layer1.1.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %71 : Tensor = aten::add(%out.3, %55, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.1 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %72 : Tensor = aten::relu(%71), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.1/__module.kp_detector.fg_encoder.layer1.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0, Tactic: 0x30e8a8d7a953e5e9, (Unnamed Layer* 24) [Activation]_output (Half[1,64:8,64,64]), (Unnamed Layer* 21) [Activation]_output (Half[1,64:8,64,64]) -> (Unnamed Layer* 28) [Activation]_output (Half[1,64:8,64,64])
Layer(CaskConvolution): %input.35 : Tensor = aten::_convolution(%72, %self.kp_detector.fg_encoder.layer2.0.conv1.weight, %5, %25, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.0/__module.kp_detector.fg_encoder.layer2.0.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.37 : Tensor = aten::batch_norm(%input.35, %self.kp_detector.fg_encoder.layer2.0.bn1.weight, %self.kp_detector.fg_encoder.layer2.0.bn1.bias, %self.kp_detector.fg_encoder.layer2.0.bn1.running_mean, %self.kp_detector.fg_encoder.layer2.0.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.0/__module.kp_detector.fg_encoder.layer2.0.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %80 : Tensor = aten::relu(%input.37), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.0/__module.kp_detector.fg_encoder.layer2.0.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0, Tactic: 0xe1ff5ad20f5c6bf6, (Unnamed Layer* 28) [Activation]_output (Half[1,64:8,64,64]) -> (Unnamed Layer* 31) [Activation]_output (Half[1,128:8,32,32])
Layer(CaskConvolution): %input.41 : Tensor = aten::_convolution(%80, %self.kp_detector.fg_encoder.layer2.0.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.0/__module.kp_detector.fg_encoder.layer2.0.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.5 : Tensor = aten::batch_norm(%input.41, %self.kp_detector.fg_encoder.layer2.0.bn2.weight, %self.kp_detector.fg_encoder.layer2.0.bn2.bias, %self.kp_detector.fg_encoder.layer2.0.bn2.running_mean, %self.kp_detector.fg_encoder.layer2.0.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.0/__module.kp_detector.fg_encoder.layer2.0.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0, Tactic: 0xe1ff5ad20f5c6bf6, (Unnamed Layer* 31) [Activation]_output (Half[1,128:8,32,32]) -> (Unnamed Layer* 35) [Scale]_output (Half[1,128:8,32,32])
Layer(CaskConvolution): %input.43 : Tensor = aten::_convolution(%72, %self.kp_detector.fg_encoder.layer2.0.downsample.0.weight, %5, %25, %28, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.0/__module.kp_detector.fg_encoder.layer2.0.downsample/__module.kp_detector.fg_encoder.layer2.0.downsample.0 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %identity.1 : Tensor = aten::batch_norm(%input.43, %self.kp_detector.fg_encoder.layer2.0.downsample.1.weight, %self.kp_detector.fg_encoder.layer2.0.bn2.bias, %self.kp_detector.fg_encoder.layer2.0.downsample.1.running_mean, %self.kp_detector.fg_encoder.layer2.0.downsample.1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.0/__module.kp_detector.fg_encoder.layer2.0.downsample/__module.kp_detector.fg_encoder.layer2.0.downsample.1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %94 : Tensor = aten::add(%out.5, %identity.1, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.0 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %95 : Tensor = aten::relu(%94), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.0/__module.kp_detector.fg_encoder.layer2.0.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0, Tactic: 0x2aa016c86360697f, (Unnamed Layer* 28) [Activation]_output (Half[1,64:8,64,64]), (Unnamed Layer* 35) [Scale]_output (Half[1,128:8,32,32]) -> (Unnamed Layer* 37) [Activation]_output (Half[1,128:8,32,32])
Layer(CaskConvolution): %input.49 : Tensor = aten::_convolution(%95, %self.kp_detector.fg_encoder.layer2.1.conv1.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.1/__module.kp_detector.fg_encoder.layer2.1.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.51 : Tensor = aten::batch_norm(%input.49, %self.kp_detector.fg_encoder.layer2.1.bn1.weight, %self.kp_detector.fg_encoder.layer2.1.bn1.bias, %self.kp_detector.fg_encoder.layer2.1.bn1.running_mean, %self.kp_detector.fg_encoder.layer2.1.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.1/__module.kp_detector.fg_encoder.layer2.1.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %103 : Tensor = aten::relu(%input.51), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.1/__module.kp_detector.fg_encoder.layer2.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0, Tactic: 0xe1ff5ad20f5c6bf6, (Unnamed Layer* 37) [Activation]_output (Half[1,128:8,32,32]) -> (Unnamed Layer* 40) [Activation]_output (Half[1,128:8,32,32])
Layer(CaskConvolution): %input.55 : Tensor = aten::_convolution(%103, %self.kp_detector.fg_encoder.layer2.1.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.1/__module.kp_detector.fg_encoder.layer2.1.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.7 : Tensor = aten::batch_norm(%input.55, %self.kp_detector.fg_encoder.layer2.1.bn2.weight, %self.kp_detector.fg_encoder.layer2.1.bn2.bias, %self.kp_detector.fg_encoder.layer2.1.bn2.running_mean, %self.kp_detector.fg_encoder.layer2.1.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.1/__module.kp_detector.fg_encoder.layer2.1.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %111 : Tensor = aten::add(%out.7, %95, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.1 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %112 : Tensor = aten::relu(%111), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.1/__module.kp_detector.fg_encoder.layer2.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0, Tactic: 0xe1ff5ad20f5c6bf6, (Unnamed Layer* 40) [Activation]_output (Half[1,128:8,32,32]), (Unnamed Layer* 37) [Activation]_output (Half[1,128:8,32,32]) -> (Unnamed Layer* 44) [Activation]_output (Half[1,128:8,32,32])
Layer(CaskConvolution): %input.61 : Tensor = aten::_convolution(%112, %self.kp_detector.fg_encoder.layer3.0.conv1.weight, %5, %25, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.63 : Tensor = aten::batch_norm(%input.61, %self.kp_detector.fg_encoder.layer3.0.bn1.weight, %self.kp_detector.fg_encoder.layer3.0.bn1.bias, %self.kp_detector.fg_encoder.layer3.0.bn1.running_mean, %self.kp_detector.fg_encoder.layer3.0.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %120 : Tensor = aten::relu(%input.63), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0, Tactic: 0xa033e20ae9f412b2, (Unnamed Layer* 44) [Activation]_output (Half[1,128:8,32,32]) -> (Unnamed Layer* 47) [Activation]_output (Half[1,256:8,16,16])
Layer(CaskConvolution): %input.67 : Tensor = aten::_convolution(%120, %self.kp_detector.fg_encoder.layer3.0.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.9 : Tensor = aten::batch_norm(%input.67, %self.kp_detector.fg_encoder.layer3.0.bn2.weight, %self.kp_detector.fg_encoder.layer3.0.bn2.bias, %self.kp_detector.fg_encoder.layer3.0.bn2.running_mean, %self.kp_detector.fg_encoder.layer3.0.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0, Tactic: 0xa033e20ae9f412b2, (Unnamed Layer* 47) [Activation]_output (Half[1,256:8,16,16]) -> (Unnamed Layer* 51) [Scale]_output (Half[1,256:8,16,16])
Layer(CaskConvolution): %input.69 : Tensor = aten::_convolution(%112, %self.kp_detector.fg_encoder.layer3.0.downsample.0.weight, %5, %25, %28, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.downsample/__module.kp_detector.fg_encoder.layer3.0.downsample.0 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %identity.3 : Tensor = aten::batch_norm(%input.69, %self.kp_detector.fg_encoder.layer3.0.downsample.1.weight, %self.kp_detector.fg_encoder.layer3.0.bn2.bias, %self.kp_detector.fg_encoder.layer3.0.downsample.1.running_mean, %self.kp_detector.fg_encoder.layer3.0.downsample.1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.downsample/__module.kp_detector.fg_encoder.layer3.0.downsample.1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %134 : Tensor = aten::add(%out.9, %identity.3, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %135 : Tensor = aten::relu(%134), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0, Tactic: 0x2aa016c86360697f, (Unnamed Layer* 44) [Activation]_output (Half[1,128:8,32,32]), (Unnamed Layer* 51) [Scale]_output (Half[1,256:8,16,16]) -> (Unnamed Layer* 53) [Activation]_output (Half[1,256:8,16,16])
Layer(CaskConvolution): %input.75 : Tensor = aten::_convolution(%135, %self.kp_detector.fg_encoder.layer3.1.conv1.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1/__module.kp_detector.fg_encoder.layer3.1.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.77 : Tensor = aten::batch_norm(%input.75, %self.kp_detector.fg_encoder.layer3.1.bn1.weight, %self.kp_detector.fg_encoder.layer3.1.bn1.bias, %self.kp_detector.fg_encoder.layer3.1.bn1.running_mean, %self.kp_detector.fg_encoder.layer3.1.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1/__module.kp_detector.fg_encoder.layer3.1.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %143 : Tensor = aten::relu(%input.77), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1/__module.kp_detector.fg_encoder.layer3.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0, Tactic: 0xa033e20ae9f412b2, (Unnamed Layer* 53) [Activation]_output (Half[1,256:8,16,16]) -> (Unnamed Layer* 56) [Activation]_output (Half[1,256:8,16,16])
Layer(CaskConvolution): %input.81 : Tensor = aten::_convolution(%143, %self.kp_detector.fg_encoder.layer3.1.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1/__module.kp_detector.fg_encoder.layer3.1.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.11 : Tensor = aten::batch_norm(%input.81, %self.kp_detector.fg_encoder.layer3.1.bn2.weight, %self.kp_detector.fg_encoder.layer3.1.bn2.bias, %self.kp_detector.fg_encoder.layer3.1.bn2.running_mean, %self.kp_detector.fg_encoder.layer3.1.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1/__module.kp_detector.fg_encoder.layer3.1.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %151 : Tensor = aten::add(%out.11, %135, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %152 : Tensor = aten::relu(%151), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1/__module.kp_detector.fg_encoder.layer3.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0, Tactic: 0xa033e20ae9f412b2, (Unnamed Layer* 56) [Activation]_output (Half[1,256:8,16,16]), (Unnamed Layer* 53) [Activation]_output (Half[1,256:8,16,16]) -> (Unnamed Layer* 60) [Activation]_output (Half[1,256:8,16,16])
Layer(CaskConvolution): %input.87 : Tensor = aten::_convolution(%152, %self.kp_detector.fg_encoder.layer4.0.conv1.weight, %5, %25, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.89 : Tensor = aten::batch_norm(%input.87, %self.kp_detector.fg_encoder.layer4.0.bn1.weight, %self.kp_detector.fg_encoder.layer4.0.bn1.bias, %self.kp_detector.fg_encoder.layer4.0.bn1.running_mean, %self.kp_detector.fg_encoder.layer4.0.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %160 : Tensor = aten::relu(%input.89), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0, Tactic: 0xa033e20ae9f412b2, (Unnamed Layer* 60) [Activation]_output (Half[1,256:8,16,16]) -> (Unnamed Layer* 63) [Activation]_output (Half[1,512:8,8,8])
Layer(CaskConvolution): %input.93 : Tensor = aten::_convolution(%160, %self.kp_detector.fg_encoder.layer4.0.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.13 : Tensor = aten::batch_norm(%input.93, %self.kp_detector.fg_encoder.layer4.0.bn2.weight, %self.kp_detector.fg_encoder.layer4.0.bn2.bias, %self.kp_detector.fg_encoder.layer4.0.bn2.running_mean, %self.kp_detector.fg_encoder.layer4.0.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0, Tactic: 0xa033e20ae9f412b2, (Unnamed Layer* 63) [Activation]_output (Half[1,512:8,8,8]) -> (Unnamed Layer* 67) [Scale]_output (Half[1,512:8,8,8])
Layer(CaskConvolution): %input.95 : Tensor = aten::_convolution(%152, %self.kp_detector.fg_encoder.layer4.0.downsample.0.weight, %5, %25, %28, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.downsample/__module.kp_detector.fg_encoder.layer4.0.downsample.0 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %identity.5 : Tensor = aten::batch_norm(%input.95, %self.kp_detector.fg_encoder.layer4.0.downsample.1.weight, %self.kp_detector.fg_encoder.layer4.0.bn2.bias, %self.kp_detector.fg_encoder.layer4.0.downsample.1.running_mean, %self.kp_detector.fg_encoder.layer4.0.downsample.1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.downsample/__module.kp_detector.fg_encoder.layer4.0.downsample.1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %174 : Tensor = aten::add(%out.13, %identity.5, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %175 : Tensor = aten::relu(%174), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0, Tactic: 0x2aa016c86360697f, (Unnamed Layer* 60) [Activation]_output (Half[1,256:8,16,16]), (Unnamed Layer* 67) [Scale]_output (Half[1,512:8,8,8]) -> (Unnamed Layer* 69) [Activation]_output (Half[1,512:8,8,8])
Layer(CaskConvolution): %input.101 : Tensor = aten::_convolution(%175, %self.kp_detector.fg_encoder.layer4.1.conv1.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1/__module.kp_detector.fg_encoder.layer4.1.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.103 : Tensor = aten::batch_norm(%input.101, %self.kp_detector.fg_encoder.layer4.1.bn1.weight, %self.kp_detector.fg_encoder.layer4.1.bn1.bias, %self.kp_detector.fg_encoder.layer4.1.bn1.running_mean, %self.kp_detector.fg_encoder.layer4.1.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1/__module.kp_detector.fg_encoder.layer4.1.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %183 : Tensor = aten::relu(%input.103), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1/__module.kp_detector.fg_encoder.layer4.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0, Tactic: 0xa033e20ae9f412b2, (Unnamed Layer* 69) [Activation]_output (Half[1,512:8,8,8]) -> (Unnamed Layer* 72) [Activation]_output (Half[1,512:8,8,8])
Layer(CaskConvolution): %input.107 : Tensor = aten::_convolution(%183, %self.kp_detector.fg_encoder.layer4.1.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1/__module.kp_detector.fg_encoder.layer4.1.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.15 : Tensor = aten::batch_norm(%input.107, %self.kp_detector.fg_encoder.layer4.1.bn2.weight, %self.kp_detector.fg_encoder.layer4.1.bn2.bias, %self.kp_detector.fg_encoder.layer4.1.bn2.running_mean, %self.kp_detector.fg_encoder.layer4.1.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1/__module.kp_detector.fg_encoder.layer4.1.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %191 : Tensor = aten::add(%out.15, %175, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %192 : Tensor = aten::relu(%191), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1/__module.kp_detector.fg_encoder.layer4.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0, Tactic: 0xa033e20ae9f412b2, (Unnamed Layer* 72) [Activation]_output (Half[1,512:8,8,8]), (Unnamed Layer* 69) [Activation]_output (Half[1,512:8,8,8]) -> (Unnamed Layer* 76) [Activation]_output (Half[1,512:8,8,8])
Layer(CaskPooling): %x.5 : Tensor = aten::adaptive_avg_pool2d(%192, %27), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.avgpool # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1213:0, Tactic: 0x56d7b61f084f251e, (Unnamed Layer* 76) [Activation]_output (Half[1,512:8,8,8]) -> (Unnamed Layer* 77) [Reduce]_output (Half[1,512:8,1,1])
Layer(NoOp): Reformatting CopyNode for Input Tensor 0 to %198 : Tensor = aten::matmul(%input.113, %196) + [Freeze Tensor %199 : Tensor = trt::const(%self.kp_detector.fg_encoder.fc.bias) ] + (Unnamed Layer* 83) [Shuffle] + unsqueeze_node_after_[Freeze Tensor %199 : Tensor = trt::const(%self.kp_detector.fg_encoder.fc.bias) ] + (Unnamed Layer* 83) [Shuffle]_(Unnamed Layer* 83) [Shuffle]_output + %201 : Tensor = aten::add(%199, %198, %365) + PWN(%fg_kp.3 : Tensor = aten::sigmoid(%201), scope: __module.kp_detector # /home/joy/remote/stable-diffusion-backend/personalize/thin_plate_spline_motion_model/tpsnn.py:43:0), Tactic: 0x0000000000000000, (Unnamed Layer* 77) [Reduce]_output (Half[1,512:8,1,1]) -> Reformatted Input Tensor 0 to %198 : Tensor = aten::matmul(%input.113, %196) + [Freeze Tensor %199 : Tensor = trt::const(%self.kp_detector.fg_encoder.fc.bias) ] + (Unnamed Layer* 83) [Shuffle] + unsqueeze_node_after_[Freeze Tensor %199 : Tensor = trt::const(%self.kp_detector.fg_encoder.fc.bias) ] + (Unnamed Layer* 83) [Shuffle]_(Unnamed Layer* 83) [Shuffle]_output + %201 : Tensor = aten::add(%199, %198, %365) + PWN(%fg_kp.3 : Tensor = aten::sigmoid(%201), scope: __module.kp_detector # /home/joy/remote/stable-diffusion-backend/personalize/thin_plate_spline_motion_model/tpsnn.py:43:0) (Half[1,512,1,1])
Layer(CublasConvolution): %198 : Tensor = aten::matmul(%input.113, %196) + [Freeze Tensor %199 : Tensor = trt::const(%self.kp_detector.fg_encoder.fc.bias) ] + (Unnamed Layer* 83) [Shuffle] + unsqueeze_node_after_[Freeze Tensor %199 : Tensor = trt::const(%self.kp_detector.fg_encoder.fc.bias) ] + (Unnamed Layer* 83) [Shuffle]_(Unnamed Layer* 83) [Shuffle]_output + %201 : Tensor = aten::add(%199, %198, %365) + PWN(%fg_kp.3 : Tensor = aten::sigmoid(%201), scope: __module.kp_detector # /home/joy/remote/stable-diffusion-backend/personalize/thin_plate_spline_motion_model/tpsnn.py:43:0), Tactic: 0x0000000000000002, Reformatted Input Tensor 0 to %198 : Tensor = aten::matmul(%input.113, %196) + [Freeze Tensor %199 : Tensor = trt::const(%self.kp_detector.fg_encoder.fc.bias) ] + (Unnamed Layer* 83) [Shuffle] + unsqueeze_node_after_[Freeze Tensor %199 : Tensor = trt::const(%self.kp_detector.fg_encoder.fc.bias) ] + (Unnamed Layer* 83) [Shuffle]_(Unnamed Layer* 83) [Shuffle]_output + %201 : Tensor = aten::add(%199, %198, %365) + PWN(%fg_kp.3 : Tensor = aten::sigmoid(%201), scope: __module.kp_detector # /home/joy/remote/stable-diffusion-backend/personalize/thin_plate_spline_motion_model/tpsnn.py:43:0) (Half[1,512,1,1]) -> %fg_kp.3 : Tensor = aten::sigmoid(%201), scope: __module.kp_detector # /home/joy/remote/stable-diffusion-backend/personalize/thin_plate_spline_motion_model/tpsnn.py:43:0_out_tensor (Half[1,100,1,1])
Layer(Reformat): Reformatting CopyNode for Input Tensor 0 to %input.231 : Tensor = aten::_convolution(%22, %self.kp_detector.fg_encoder.conv1.weight, %5, %25, %26, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.233 : Tensor = aten::batch_norm(%input.231, %self.kp_detector.fg_encoder.bn1.weight, %self.kp_detector.fg_encoder.bn1.bias, %self.kp_detector.fg_encoder.bn1.running_mean, %self.kp_detector.fg_encoder.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %295 : Tensor = aten::relu(%input.233), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 || %input.117 : Tensor = aten::_convolution(%input.115, %self.kp_detector.fg_encoder.conv1.weight, %5, %25, %26, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.119 : Tensor = aten::batch_norm(%input.117, %self.kp_detector.fg_encoder.bn1.weight, %self.kp_detector.fg_encoder.bn1.bias, %self.kp_detector.fg_encoder.bn1.running_mean, %self.kp_detector.fg_encoder.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %216 : Tensor = aten::relu(%input.119), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0, Tactic: 0x0000000000000000, (Unnamed Layer* 10) [Shuffle]_output (Float[1,3,256,256]) -> Reformatted Input Tensor 0 to %input.231 : Tensor = aten::_convolution(%22, %self.kp_detector.fg_encoder.conv1.weight, %5, %25, %26, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.233 : Tensor = aten::batch_norm(%input.231, %self.kp_detector.fg_encoder.bn1.weight, %self.kp_detector.fg_encoder.bn1.bias, %self.kp_detector.fg_encoder.bn1.running_mean, %self.kp_detector.fg_encoder.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %295 : Tensor = aten::relu(%input.233), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 || %input.117 : Tensor = aten::_convolution(%input.115, %self.kp_detector.fg_encoder.conv1.weight, %5, %25, %26, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.119 : Tensor = aten::batch_norm(%input.117, %self.kp_detector.fg_encoder.bn1.weight, %self.kp_detector.fg_encoder.bn1.bias, %self.kp_detector.fg_encoder.bn1.running_mean, %self.kp_detector.fg_encoder.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %216 : Tensor = aten::relu(%input.119), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 (Half[1,3:2,256,256])
Layer(CaskConvolution): %input.231 : Tensor = aten::_convolution(%22, %self.kp_detector.fg_encoder.conv1.weight, %5, %25, %26, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.233 : Tensor = aten::batch_norm(%input.231, %self.kp_detector.fg_encoder.bn1.weight, %self.kp_detector.fg_encoder.bn1.bias, %self.kp_detector.fg_encoder.bn1.running_mean, %self.kp_detector.fg_encoder.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %295 : Tensor = aten::relu(%input.233), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 || %input.117 : Tensor = aten::_convolution(%input.115, %self.kp_detector.fg_encoder.conv1.weight, %5, %25, %26, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.119 : Tensor = aten::batch_norm(%input.117, %self.kp_detector.fg_encoder.bn1.weight, %self.kp_detector.fg_encoder.bn1.bias, %self.kp_detector.fg_encoder.bn1.running_mean, %self.kp_detector.fg_encoder.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %216 : Tensor = aten::relu(%input.119), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0, Tactic: 0x4cfee77ea8c324db, Reformatted Input Tensor 0 to %input.231 : Tensor = aten::_convolution(%22, %self.kp_detector.fg_encoder.conv1.weight, %5, %25, %26, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.233 : Tensor = aten::batch_norm(%input.231, %self.kp_detector.fg_encoder.bn1.weight, %self.kp_detector.fg_encoder.bn1.bias, %self.kp_detector.fg_encoder.bn1.running_mean, %self.kp_detector.fg_encoder.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %295 : Tensor = aten::relu(%input.233), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 || %input.117 : Tensor = aten::_convolution(%input.115, %self.kp_detector.fg_encoder.conv1.weight, %5, %25, %26, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.119 : Tensor = aten::batch_norm(%input.117, %self.kp_detector.fg_encoder.bn1.weight, %self.kp_detector.fg_encoder.bn1.bias, %self.kp_detector.fg_encoder.bn1.running_mean, %self.kp_detector.fg_encoder.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %216 : Tensor = aten::relu(%input.119), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 (Half[1,3:2,256,256]) -> %input.231 : Tensor = aten::_convolution(%22, %self.kp_detector.fg_encoder.conv1.weight, %5, %25, %26, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.233 : Tensor = aten::batch_norm(%input.231, %self.kp_detector.fg_encoder.bn1.weight, %self.kp_detector.fg_encoder.bn1.bias, %self.kp_detector.fg_encoder.bn1.running_mean, %self.kp_detector.fg_encoder.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %295 : Tensor = aten::relu(%input.233), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 || %input.117 : Tensor = aten::_convolution(%input.115, %self.kp_detector.fg_encoder.conv1.weight, %5, %25, %26, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.119 : Tensor = aten::batch_norm(%input.117, %self.kp_detector.fg_encoder.bn1.weight, %self.kp_detector.fg_encoder.bn1.bias, %self.kp_detector.fg_encoder.bn1.running_mean, %self.kp_detector.fg_encoder.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %216 : Tensor = aten::relu(%input.119), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 (Half[1,128:2,128,128])
Layer(ElementWise): %205 : Tensor = aten::mul(%fg_kp.3, %206), scope: __module.kp_detector # /home/joy/remote/stable-diffusion-backend/personalize/thin_plate_spline_motion_model/tpsnn.py:43:0, Tactic: 0x0000000000000001, %fg_kp.3 : Tensor = aten::sigmoid(%201), scope: __module.kp_detector # /home/joy/remote/stable-diffusion-backend/personalize/thin_plate_spline_motion_model/tpsnn.py:43:0_out_tensor (Half[1,100,1,1]), unsqueeze_tensor_after_{ForeignNode[(Unnamed Layer* 1) [Shuffle]...%input.115 : Tensor = aten::select(%19, %21, %16) # /home/joy/remote/stable-diffusion-backend/personalize/thin_plate_spline_motion_model/tpsnn.py:529:0]}_(Unnamed Layer* 88) [Shuffle]_output_out_tensor (Half[1,1,1,1]) -> %205 : Tensor = aten::mul(%fg_kp.3, %206), scope: __module.kp_detector # /home/joy/remote/stable-diffusion-backend/personalize/thin_plate_spline_motion_model/tpsnn.py:43:0_out_tensor (Half[1,100,1,1])
Layer(NoOp): copied_squeeze_after_%205 : Tensor = aten::mul(%fg_kp.3, %206), scope: __module.kp_detector # /home/joy/remote/stable-diffusion-backend/personalize/thin_plate_spline_motion_model/tpsnn.py:43:0, Tactic: 0x0000000000000000, %205 : Tensor = aten::mul(%fg_kp.3, %206), scope: __module.kp_detector # /home/joy/remote/stable-diffusion-backend/personalize/thin_plate_spline_motion_model/tpsnn.py:43:0_out_tensor (Half[1,100,1,1]) -> (Unnamed Layer* 89) [ElementWise]_output (Half[1,100])
Layer(Reformat): Reformatting CopyNode for Input Tensor 0 to %fg_kp.5 : Tensor = aten::sub(%205, %208, %365), scope: __module.kp_detector # /home/joy/remote/stable-diffusion-backend/personalize/thin_plate_spline_motion_model/tpsnn.py:43:0, Tactic: 0x00000000000003e8, (Unnamed Layer* 89) [ElementWise]_output (Half[1,100]) -> Reformatted Input Tensor 0 to %fg_kp.5 : Tensor = aten::sub(%205, %208, %365), scope: __module.kp_detector # /home/joy/remote/stable-diffusion-backend/personalize/thin_plate_spline_motion_model/tpsnn.py:43:0 (Float[1,100])
Layer(ElementWise): %fg_kp.5 : Tensor = aten::sub(%205, %208, %365), scope: __module.kp_detector # /home/joy/remote/stable-diffusion-backend/personalize/thin_plate_spline_motion_model/tpsnn.py:43:0, Tactic: 0x0000000000000001, Reformatted Input Tensor 0 to %fg_kp.5 : Tensor = aten::sub(%205, %208, %365), scope: __module.kp_detector # /home/joy/remote/stable-diffusion-backend/personalize/thin_plate_spline_motion_model/tpsnn.py:43:0 (Float[1,100]), (Unnamed Layer* 92) [Shuffle]_output (Float[1,1]) -> (Unnamed Layer* 93) [ElementWise]_output (Float[1,100])
Layer(NoOp): %211 : Tensor = aten::reshape(%fg_kp.5, %209), Tactic: 0x0000000000000000, (Unnamed Layer* 93) [ElementWise]_output (Float[1,100]) -> output_2 (Float[1,50,2])
Layer(Reduce): %374 : Tensor = aten::mean(%211, %373, %4, %5) # /home/joy/remote/stable-diffusion-backend/personalize/thin_plate_spline_motion_model/tpsnn.py:503:0, Tactic: 0x0000000000000008, output_2 (Float[1,50,2]) -> (Unnamed Layer* 265) [Reduce]_output (Float[1,2])
Layer(Reformat): Reformatting CopyNode for Input Tensor 0 to %212 : Tensor = aten::reshape(%211, %213), Tactic: 0x0000000000000000, output_2 (Float[1,50,2]) -> Reformatted Input Tensor 0 to %212 : Tensor = aten::reshape(%211, %213) (Float[1,50,2])
Layer(NoOp): %212 : Tensor = aten::reshape(%211, %213), Tactic: 0x0000000000000000, Reformatted Input Tensor 0 to %212 : Tensor = aten::reshape(%211, %213) (Float[1,50,2]) -> Reformatted Output Tensor 0 to %212 : Tensor = aten::reshape(%211, %213) (Float[1,50,1,1,2])
Layer(Reformat): Reformatting CopyNode for Output Tensor 0 to %212 : Tensor = aten::reshape(%211, %213), Tactic: 0x00000000000003e8, Reformatted Output Tensor 0 to %212 : Tensor = aten::reshape(%211, %213) (Float[1,50,1,1,2]) -> output_4 (Float[1,50,1,1,2])
Layer(CaskPooling): %input.123 : Tensor = aten::max_pool2d(%216, %26, %25, %27, %27, %4), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.maxpool # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:780:0, Tactic: 0xc8ae100b63fd8921, %input.231 : Tensor = aten::_convolution(%22, %self.kp_detector.fg_encoder.conv1.weight, %5, %25, %26, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.233 : Tensor = aten::batch_norm(%input.231, %self.kp_detector.fg_encoder.bn1.weight, %self.kp_detector.fg_encoder.bn1.bias, %self.kp_detector.fg_encoder.bn1.running_mean, %self.kp_detector.fg_encoder.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %295 : Tensor = aten::relu(%input.233), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 || %input.117 : Tensor = aten::_convolution(%input.115, %self.kp_detector.fg_encoder.conv1.weight, %5, %25, %26, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.119 : Tensor = aten::batch_norm(%input.117, %self.kp_detector.fg_encoder.bn1.weight, %self.kp_detector.fg_encoder.bn1.bias, %self.kp_detector.fg_encoder.bn1.running_mean, %self.kp_detector.fg_encoder.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %216 : Tensor = aten::relu(%input.119), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 (Half[1,64:2,128,128]) -> Reformatted Output Tensor 0 to %input.123 : Tensor = aten::max_pool2d(%216, %26, %25, %27, %27, %4), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.maxpool # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:780:0 (Half[1,64:2,64,64])
Layer(Reformat): Reformatting CopyNode for Output Tensor 0 to %input.123 : Tensor = aten::max_pool2d(%216, %26, %25, %27, %27, %4), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.maxpool # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:780:0, Tactic: 0x0000000000000000, Reformatted Output Tensor 0 to %input.123 : Tensor = aten::max_pool2d(%216, %26, %25, %27, %27, %4), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.maxpool # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:780:0 (Half[1,64:2,64,64]) -> (Unnamed Layer* 99) [Pooling]_output (Half[1,64:8,64,64])
Layer(CaskConvolution): %input.125 : Tensor = aten::_convolution(%input.123, %self.kp_detector.fg_encoder.layer1.0.conv1.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.0/__module.kp_detector.fg_encoder.layer1.0.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.127 : Tensor = aten::batch_norm(%input.125, %self.kp_detector.fg_encoder.layer1.0.bn1.weight, %self.kp_detector.fg_encoder.layer1.0.bn1.bias, %self.kp_detector.fg_encoder.layer1.0.bn1.running_mean, %self.kp_detector.fg_encoder.layer1.0.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.0/__module.kp_detector.fg_encoder.layer1.0.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %220 : Tensor = aten::relu(%input.127), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.0/__module.kp_detector.fg_encoder.layer1.0.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0, Tactic: 0x30e8a8d7a953e5e9, (Unnamed Layer* 99) [Pooling]_output (Half[1,64:8,64,64]) -> (Unnamed Layer* 102) [Activation]_output (Half[1,64:8,64,64])
Layer(CaskPooling): %input.237 : Tensor = aten::max_pool2d(%295, %26, %25, %27, %27, %4), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.maxpool # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:780:0, Tactic: 0xc8ae100b63fd8921, %input.231 : Tensor = aten::_convolution(%22, %self.kp_detector.fg_encoder.conv1.weight, %5, %25, %26, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.233 : Tensor = aten::batch_norm(%input.231, %self.kp_detector.fg_encoder.bn1.weight, %self.kp_detector.fg_encoder.bn1.bias, %self.kp_detector.fg_encoder.bn1.running_mean, %self.kp_detector.fg_encoder.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %295 : Tensor = aten::relu(%input.233), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 || %input.117 : Tensor = aten::_convolution(%input.115, %self.kp_detector.fg_encoder.conv1.weight, %5, %25, %26, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.119 : Tensor = aten::batch_norm(%input.117, %self.kp_detector.fg_encoder.bn1.weight, %self.kp_detector.fg_encoder.bn1.bias, %self.kp_detector.fg_encoder.bn1.running_mean, %self.kp_detector.fg_encoder.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %216 : Tensor = aten::relu(%input.119), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0 (Half[1,64:2,128,128]) -> Reformatted Output Tensor 0 to %input.237 : Tensor = aten::max_pool2d(%295, %26, %25, %27, %27, %4), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.maxpool # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:780:0 (Half[1,64:2,64,64])
Layer(Reformat): Reformatting CopyNode for Output Tensor 0 to %input.237 : Tensor = aten::max_pool2d(%295, %26, %25, %27, %27, %4), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.maxpool # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:780:0, Tactic: 0x0000000000000000, Reformatted Output Tensor 0 to %input.237 : Tensor = aten::max_pool2d(%295, %26, %25, %27, %27, %4), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.maxpool # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:780:0 (Half[1,64:2,64,64]) -> (Unnamed Layer* 183) [Pooling]_output (Half[1,64:8,64,64])
Layer(CaskConvolution): %input.239 : Tensor = aten::_convolution(%input.237, %self.kp_detector.fg_encoder.layer1.0.conv1.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.0/__module.kp_detector.fg_encoder.layer1.0.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.241 : Tensor = aten::batch_norm(%input.239, %self.kp_detector.fg_encoder.layer1.0.bn1.weight, %self.kp_detector.fg_encoder.layer1.0.bn1.bias, %self.kp_detector.fg_encoder.layer1.0.bn1.running_mean, %self.kp_detector.fg_encoder.layer1.0.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.0/__module.kp_detector.fg_encoder.layer1.0.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %299 : Tensor = aten::relu(%input.241), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.0/__module.kp_detector.fg_encoder.layer1.0.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0, Tactic: 0x30e8a8d7a953e5e9, (Unnamed Layer* 183) [Pooling]_output (Half[1,64:8,64,64]) -> (Unnamed Layer* 186) [Activation]_output (Half[1,64:8,64,64])
Layer(CaskConvolution): %input.245 : Tensor = aten::_convolution(%299, %self.kp_detector.fg_encoder.layer1.0.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.0/__module.kp_detector.fg_encoder.layer1.0.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.33 : Tensor = aten::batch_norm(%input.245, %self.kp_detector.fg_encoder.layer1.0.bn2.weight, %self.kp_detector.fg_encoder.layer1.0.bn2.bias, %self.kp_detector.fg_encoder.layer1.0.bn2.running_mean, %self.kp_detector.fg_encoder.layer1.0.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.0/__module.kp_detector.fg_encoder.layer1.0.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %302 : Tensor = aten::add(%out.33, %input.237, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.0 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %303 : Tensor = aten::relu(%302), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.0/__module.kp_detector.fg_encoder.layer1.0.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0, Tactic: 0x30e8a8d7a953e5e9, (Unnamed Layer* 186) [Activation]_output (Half[1,64:8,64,64]), (Unnamed Layer* 183) [Pooling]_output (Half[1,64:8,64,64]) -> (Unnamed Layer* 190) [Activation]_output (Half[1,64:8,64,64])
Layer(CaskConvolution): %input.131 : Tensor = aten::_convolution(%220, %self.kp_detector.fg_encoder.layer1.0.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.0/__module.kp_detector.fg_encoder.layer1.0.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.17 : Tensor = aten::batch_norm(%input.131, %self.kp_detector.fg_encoder.layer1.0.bn2.weight, %self.kp_detector.fg_encoder.layer1.0.bn2.bias, %self.kp_detector.fg_encoder.layer1.0.bn2.running_mean, %self.kp_detector.fg_encoder.layer1.0.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.0/__module.kp_detector.fg_encoder.layer1.0.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %223 : Tensor = aten::add(%out.17, %input.123, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.0 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %224 : Tensor = aten::relu(%223), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.0/__module.kp_detector.fg_encoder.layer1.0.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0, Tactic: 0x30e8a8d7a953e5e9, (Unnamed Layer* 102) [Activation]_output (Half[1,64:8,64,64]), (Unnamed Layer* 99) [Pooling]_output (Half[1,64:8,64,64]) -> (Unnamed Layer* 106) [Activation]_output (Half[1,64:8,64,64])
Layer(CaskConvolution): %input.137 : Tensor = aten::_convolution(%224, %self.kp_detector.fg_encoder.layer1.1.conv1.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.1/__module.kp_detector.fg_encoder.layer1.1.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.139 : Tensor = aten::batch_norm(%input.137, %self.kp_detector.fg_encoder.layer1.1.bn1.weight, %self.kp_detector.fg_encoder.layer1.1.bn1.bias, %self.kp_detector.fg_encoder.layer1.1.bn1.running_mean, %self.kp_detector.fg_encoder.layer1.1.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.1/__module.kp_detector.fg_encoder.layer1.1.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %227 : Tensor = aten::relu(%input.139), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.1/__module.kp_detector.fg_encoder.layer1.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0, Tactic: 0x30e8a8d7a953e5e9, (Unnamed Layer* 106) [Activation]_output (Half[1,64:8,64,64]) -> (Unnamed Layer* 109) [Activation]_output (Half[1,64:8,64,64])
Layer(CaskConvolution): %input.251 : Tensor = aten::_convolution(%303, %self.kp_detector.fg_encoder.layer1.1.conv1.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.1/__module.kp_detector.fg_encoder.layer1.1.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.253 : Tensor = aten::batch_norm(%input.251, %self.kp_detector.fg_encoder.layer1.1.bn1.weight, %self.kp_detector.fg_encoder.layer1.1.bn1.bias, %self.kp_detector.fg_encoder.layer1.1.bn1.running_mean, %self.kp_detector.fg_encoder.layer1.1.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.1/__module.kp_detector.fg_encoder.layer1.1.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %306 : Tensor = aten::relu(%input.253), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.1/__module.kp_detector.fg_encoder.layer1.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0, Tactic: 0x30e8a8d7a953e5e9, (Unnamed Layer* 190) [Activation]_output (Half[1,64:8,64,64]) -> (Unnamed Layer* 193) [Activation]_output (Half[1,64:8,64,64])
Layer(CaskConvolution): %input.257 : Tensor = aten::_convolution(%306, %self.kp_detector.fg_encoder.layer1.1.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.1/__module.kp_detector.fg_encoder.layer1.1.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.35 : Tensor = aten::batch_norm(%input.257, %self.kp_detector.fg_encoder.layer1.1.bn2.weight, %self.kp_detector.fg_encoder.layer1.1.bn2.bias, %self.kp_detector.fg_encoder.layer1.1.bn2.running_mean, %self.kp_detector.fg_encoder.layer1.1.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.1/__module.kp_detector.fg_encoder.layer1.1.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %309 : Tensor = aten::add(%out.35, %303, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.1 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %310 : Tensor = aten::relu(%309), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.1/__module.kp_detector.fg_encoder.layer1.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0, Tactic: 0x30e8a8d7a953e5e9, (Unnamed Layer* 193) [Activation]_output (Half[1,64:8,64,64]), (Unnamed Layer* 190) [Activation]_output (Half[1,64:8,64,64]) -> (Unnamed Layer* 197) [Activation]_output (Half[1,64:8,64,64])
Layer(CaskConvolution): %input.143 : Tensor = aten::_convolution(%227, %self.kp_detector.fg_encoder.layer1.1.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.1/__module.kp_detector.fg_encoder.layer1.1.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.19 : Tensor = aten::batch_norm(%input.143, %self.kp_detector.fg_encoder.layer1.1.bn2.weight, %self.kp_detector.fg_encoder.layer1.1.bn2.bias, %self.kp_detector.fg_encoder.layer1.1.bn2.running_mean, %self.kp_detector.fg_encoder.layer1.1.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.1/__module.kp_detector.fg_encoder.layer1.1.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %230 : Tensor = aten::add(%out.19, %224, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.1 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %231 : Tensor = aten::relu(%230), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer1/__module.kp_detector.fg_encoder.layer1.1/__module.kp_detector.fg_encoder.layer1.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0, Tactic: 0x30e8a8d7a953e5e9, (Unnamed Layer* 109) [Activation]_output (Half[1,64:8,64,64]), (Unnamed Layer* 106) [Activation]_output (Half[1,64:8,64,64]) -> (Unnamed Layer* 113) [Activation]_output (Half[1,64:8,64,64])
Layer(CaskConvolution): %input.149 : Tensor = aten::_convolution(%231, %self.kp_detector.fg_encoder.layer2.0.conv1.weight, %5, %25, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.0/__module.kp_detector.fg_encoder.layer2.0.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.151 : Tensor = aten::batch_norm(%input.149, %self.kp_detector.fg_encoder.layer2.0.bn1.weight, %self.kp_detector.fg_encoder.layer2.0.bn1.bias, %self.kp_detector.fg_encoder.layer2.0.bn1.running_mean, %self.kp_detector.fg_encoder.layer2.0.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.0/__module.kp_detector.fg_encoder.layer2.0.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %234 : Tensor = aten::relu(%input.151), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.0/__module.kp_detector.fg_encoder.layer2.0.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0, Tactic: 0xe1ff5ad20f5c6bf6, (Unnamed Layer* 113) [Activation]_output (Half[1,64:8,64,64]) -> (Unnamed Layer* 116) [Activation]_output (Half[1,128:8,32,32])
Layer(CaskConvolution): %input.263 : Tensor = aten::_convolution(%310, %self.kp_detector.fg_encoder.layer2.0.conv1.weight, %5, %25, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.0/__module.kp_detector.fg_encoder.layer2.0.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.265 : Tensor = aten::batch_norm(%input.263, %self.kp_detector.fg_encoder.layer2.0.bn1.weight, %self.kp_detector.fg_encoder.layer2.0.bn1.bias, %self.kp_detector.fg_encoder.layer2.0.bn1.running_mean, %self.kp_detector.fg_encoder.layer2.0.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.0/__module.kp_detector.fg_encoder.layer2.0.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %313 : Tensor = aten::relu(%input.265), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.0/__module.kp_detector.fg_encoder.layer2.0.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0, Tactic: 0xe1ff5ad20f5c6bf6, (Unnamed Layer* 197) [Activation]_output (Half[1,64:8,64,64]) -> (Unnamed Layer* 200) [Activation]_output (Half[1,128:8,32,32])
Layer(CaskConvolution): %input.269 : Tensor = aten::_convolution(%313, %self.kp_detector.fg_encoder.layer2.0.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.0/__module.kp_detector.fg_encoder.layer2.0.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.37 : Tensor = aten::batch_norm(%input.269, %self.kp_detector.fg_encoder.layer2.0.bn2.weight, %self.kp_detector.fg_encoder.layer2.0.bn2.bias, %self.kp_detector.fg_encoder.layer2.0.bn2.running_mean, %self.kp_detector.fg_encoder.layer2.0.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.0/__module.kp_detector.fg_encoder.layer2.0.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0, Tactic: 0xe1ff5ad20f5c6bf6, (Unnamed Layer* 200) [Activation]_output (Half[1,128:8,32,32]) -> (Unnamed Layer* 204) [Scale]_output (Half[1,128:8,32,32])
Layer(CaskConvolution): %input.155 : Tensor = aten::_convolution(%234, %self.kp_detector.fg_encoder.layer2.0.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.0/__module.kp_detector.fg_encoder.layer2.0.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.21 : Tensor = aten::batch_norm(%input.155, %self.kp_detector.fg_encoder.layer2.0.bn2.weight, %self.kp_detector.fg_encoder.layer2.0.bn2.bias, %self.kp_detector.fg_encoder.layer2.0.bn2.running_mean, %self.kp_detector.fg_encoder.layer2.0.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.0/__module.kp_detector.fg_encoder.layer2.0.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0, Tactic: 0xe1ff5ad20f5c6bf6, (Unnamed Layer* 116) [Activation]_output (Half[1,128:8,32,32]) -> (Unnamed Layer* 120) [Scale]_output (Half[1,128:8,32,32])
Layer(CaskConvolution): %input.271 : Tensor = aten::_convolution(%310, %self.kp_detector.fg_encoder.layer2.0.downsample.0.weight, %5, %25, %28, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.0/__module.kp_detector.fg_encoder.layer2.0.downsample/__module.kp_detector.fg_encoder.layer2.0.downsample.0 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %identity.13 : Tensor = aten::batch_norm(%input.271, %self.kp_detector.fg_encoder.layer2.0.downsample.1.weight, %self.kp_detector.fg_encoder.layer2.0.bn2.bias, %self.kp_detector.fg_encoder.layer2.0.downsample.1.running_mean, %self.kp_detector.fg_encoder.layer2.0.downsample.1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.0/__module.kp_detector.fg_encoder.layer2.0.downsample/__module.kp_detector.fg_encoder.layer2.0.downsample.1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %318 : Tensor = aten::add(%out.37, %identity.13, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.0 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %319 : Tensor = aten::relu(%318), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.0/__module.kp_detector.fg_encoder.layer2.0.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0, Tactic: 0x2aa016c86360697f, (Unnamed Layer* 197) [Activation]_output (Half[1,64:8,64,64]), (Unnamed Layer* 204) [Scale]_output (Half[1,128:8,32,32]) -> (Unnamed Layer* 206) [Activation]_output (Half[1,128:8,32,32])
Layer(CaskConvolution): %input.157 : Tensor = aten::_convolution(%231, %self.kp_detector.fg_encoder.layer2.0.downsample.0.weight, %5, %25, %28, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.0/__module.kp_detector.fg_encoder.layer2.0.downsample/__module.kp_detector.fg_encoder.layer2.0.downsample.0 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %identity.7 : Tensor = aten::batch_norm(%input.157, %self.kp_detector.fg_encoder.layer2.0.downsample.1.weight, %self.kp_detector.fg_encoder.layer2.0.bn2.bias, %self.kp_detector.fg_encoder.layer2.0.downsample.1.running_mean, %self.kp_detector.fg_encoder.layer2.0.downsample.1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.0/__module.kp_detector.fg_encoder.layer2.0.downsample/__module.kp_detector.fg_encoder.layer2.0.downsample.1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %239 : Tensor = aten::add(%out.21, %identity.7, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.0 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %240 : Tensor = aten::relu(%239), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.0/__module.kp_detector.fg_encoder.layer2.0.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0, Tactic: 0x2aa016c86360697f, (Unnamed Layer* 113) [Activation]_output (Half[1,64:8,64,64]), (Unnamed Layer* 120) [Scale]_output (Half[1,128:8,32,32]) -> (Unnamed Layer* 122) [Activation]_output (Half[1,128:8,32,32])
Layer(CaskConvolution): %input.163 : Tensor = aten::_convolution(%240, %self.kp_detector.fg_encoder.layer2.1.conv1.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.1/__module.kp_detector.fg_encoder.layer2.1.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.165 : Tensor = aten::batch_norm(%input.163, %self.kp_detector.fg_encoder.layer2.1.bn1.weight, %self.kp_detector.fg_encoder.layer2.1.bn1.bias, %self.kp_detector.fg_encoder.layer2.1.bn1.running_mean, %self.kp_detector.fg_encoder.layer2.1.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.1/__module.kp_detector.fg_encoder.layer2.1.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %243 : Tensor = aten::relu(%input.165), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.1/__module.kp_detector.fg_encoder.layer2.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0, Tactic: 0xe1ff5ad20f5c6bf6, (Unnamed Layer* 122) [Activation]_output (Half[1,128:8,32,32]) -> (Unnamed Layer* 125) [Activation]_output (Half[1,128:8,32,32])
Layer(CaskConvolution): %input.277 : Tensor = aten::_convolution(%319, %self.kp_detector.fg_encoder.layer2.1.conv1.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.1/__module.kp_detector.fg_encoder.layer2.1.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.279 : Tensor = aten::batch_norm(%input.277, %self.kp_detector.fg_encoder.layer2.1.bn1.weight, %self.kp_detector.fg_encoder.layer2.1.bn1.bias, %self.kp_detector.fg_encoder.layer2.1.bn1.running_mean, %self.kp_detector.fg_encoder.layer2.1.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.1/__module.kp_detector.fg_encoder.layer2.1.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %322 : Tensor = aten::relu(%input.279), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.1/__module.kp_detector.fg_encoder.layer2.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0, Tactic: 0xe1ff5ad20f5c6bf6, (Unnamed Layer* 206) [Activation]_output (Half[1,128:8,32,32]) -> (Unnamed Layer* 209) [Activation]_output (Half[1,128:8,32,32])
Layer(CaskConvolution): %input.283 : Tensor = aten::_convolution(%322, %self.kp_detector.fg_encoder.layer2.1.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.1/__module.kp_detector.fg_encoder.layer2.1.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.39 : Tensor = aten::batch_norm(%input.283, %self.kp_detector.fg_encoder.layer2.1.bn2.weight, %self.kp_detector.fg_encoder.layer2.1.bn2.bias, %self.kp_detector.fg_encoder.layer2.1.bn2.running_mean, %self.kp_detector.fg_encoder.layer2.1.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.1/__module.kp_detector.fg_encoder.layer2.1.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %325 : Tensor = aten::add(%out.39, %319, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.1 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %326 : Tensor = aten::relu(%325), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.1/__module.kp_detector.fg_encoder.layer2.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0, Tactic: 0xe1ff5ad20f5c6bf6, (Unnamed Layer* 209) [Activation]_output (Half[1,128:8,32,32]), (Unnamed Layer* 206) [Activation]_output (Half[1,128:8,32,32]) -> (Unnamed Layer* 213) [Activation]_output (Half[1,128:8,32,32])
Layer(CaskConvolution): %input.169 : Tensor = aten::_convolution(%243, %self.kp_detector.fg_encoder.layer2.1.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.1/__module.kp_detector.fg_encoder.layer2.1.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.23 : Tensor = aten::batch_norm(%input.169, %self.kp_detector.fg_encoder.layer2.1.bn2.weight, %self.kp_detector.fg_encoder.layer2.1.bn2.bias, %self.kp_detector.fg_encoder.layer2.1.bn2.running_mean, %self.kp_detector.fg_encoder.layer2.1.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.1/__module.kp_detector.fg_encoder.layer2.1.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %246 : Tensor = aten::add(%out.23, %240, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.1 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %247 : Tensor = aten::relu(%246), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer2/__module.kp_detector.fg_encoder.layer2.1/__module.kp_detector.fg_encoder.layer2.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0, Tactic: 0xe1ff5ad20f5c6bf6, (Unnamed Layer* 125) [Activation]_output (Half[1,128:8,32,32]), (Unnamed Layer* 122) [Activation]_output (Half[1,128:8,32,32]) -> (Unnamed Layer* 129) [Activation]_output (Half[1,128:8,32,32])
Layer(CaskConvolution): %input.175 : Tensor = aten::_convolution(%247, %self.kp_detector.fg_encoder.layer3.0.conv1.weight, %5, %25, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.177 : Tensor = aten::batch_norm(%input.175, %self.kp_detector.fg_encoder.layer3.0.bn1.weight, %self.kp_detector.fg_encoder.layer3.0.bn1.bias, %self.kp_detector.fg_encoder.layer3.0.bn1.running_mean, %self.kp_detector.fg_encoder.layer3.0.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %250 : Tensor = aten::relu(%input.177), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0, Tactic: 0xa033e20ae9f412b2, (Unnamed Layer* 129) [Activation]_output (Half[1,128:8,32,32]) -> (Unnamed Layer* 132) [Activation]_output (Half[1,256:8,16,16])
Layer(CaskConvolution): %input.289 : Tensor = aten::_convolution(%326, %self.kp_detector.fg_encoder.layer3.0.conv1.weight, %5, %25, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.291 : Tensor = aten::batch_norm(%input.289, %self.kp_detector.fg_encoder.layer3.0.bn1.weight, %self.kp_detector.fg_encoder.layer3.0.bn1.bias, %self.kp_detector.fg_encoder.layer3.0.bn1.running_mean, %self.kp_detector.fg_encoder.layer3.0.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %329 : Tensor = aten::relu(%input.291), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0, Tactic: 0xa033e20ae9f412b2, (Unnamed Layer* 213) [Activation]_output (Half[1,128:8,32,32]) -> (Unnamed Layer* 216) [Activation]_output (Half[1,256:8,16,16])
Layer(CaskConvolution): %input.295 : Tensor = aten::_convolution(%329, %self.kp_detector.fg_encoder.layer3.0.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.41 : Tensor = aten::batch_norm(%input.295, %self.kp_detector.fg_encoder.layer3.0.bn2.weight, %self.kp_detector.fg_encoder.layer3.0.bn2.bias, %self.kp_detector.fg_encoder.layer3.0.bn2.running_mean, %self.kp_detector.fg_encoder.layer3.0.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0, Tactic: 0xa033e20ae9f412b2, (Unnamed Layer* 216) [Activation]_output (Half[1,256:8,16,16]) -> (Unnamed Layer* 220) [Scale]_output (Half[1,256:8,16,16])
Layer(CaskConvolution): %input.181 : Tensor = aten::_convolution(%250, %self.kp_detector.fg_encoder.layer3.0.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.25 : Tensor = aten::batch_norm(%input.181, %self.kp_detector.fg_encoder.layer3.0.bn2.weight, %self.kp_detector.fg_encoder.layer3.0.bn2.bias, %self.kp_detector.fg_encoder.layer3.0.bn2.running_mean, %self.kp_detector.fg_encoder.layer3.0.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0, Tactic: 0xa033e20ae9f412b2, (Unnamed Layer* 132) [Activation]_output (Half[1,256:8,16,16]) -> (Unnamed Layer* 136) [Scale]_output (Half[1,256:8,16,16])
Layer(CaskConvolution): %input.297 : Tensor = aten::_convolution(%326, %self.kp_detector.fg_encoder.layer3.0.downsample.0.weight, %5, %25, %28, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.downsample/__module.kp_detector.fg_encoder.layer3.0.downsample.0 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %identity.15 : Tensor = aten::batch_norm(%input.297, %self.kp_detector.fg_encoder.layer3.0.downsample.1.weight, %self.kp_detector.fg_encoder.layer3.0.bn2.bias, %self.kp_detector.fg_encoder.layer3.0.downsample.1.running_mean, %self.kp_detector.fg_encoder.layer3.0.downsample.1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.downsample/__module.kp_detector.fg_encoder.layer3.0.downsample.1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %334 : Tensor = aten::add(%out.41, %identity.15, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %335 : Tensor = aten::relu(%334), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0, Tactic: 0x2aa016c86360697f, (Unnamed Layer* 213) [Activation]_output (Half[1,128:8,32,32]), (Unnamed Layer* 220) [Scale]_output (Half[1,256:8,16,16]) -> (Unnamed Layer* 222) [Activation]_output (Half[1,256:8,16,16])
Layer(CaskConvolution): %input.183 : Tensor = aten::_convolution(%247, %self.kp_detector.fg_encoder.layer3.0.downsample.0.weight, %5, %25, %28, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.downsample/__module.kp_detector.fg_encoder.layer3.0.downsample.0 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %identity.9 : Tensor = aten::batch_norm(%input.183, %self.kp_detector.fg_encoder.layer3.0.downsample.1.weight, %self.kp_detector.fg_encoder.layer3.0.bn2.bias, %self.kp_detector.fg_encoder.layer3.0.downsample.1.running_mean, %self.kp_detector.fg_encoder.layer3.0.downsample.1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.downsample/__module.kp_detector.fg_encoder.layer3.0.downsample.1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %255 : Tensor = aten::add(%out.25, %identity.9, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %256 : Tensor = aten::relu(%255), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.0/__module.kp_detector.fg_encoder.layer3.0.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0, Tactic: 0x2aa016c86360697f, (Unnamed Layer* 129) [Activation]_output (Half[1,128:8,32,32]), (Unnamed Layer* 136) [Scale]_output (Half[1,256:8,16,16]) -> (Unnamed Layer* 138) [Activation]_output (Half[1,256:8,16,16])
Layer(CaskConvolution): %input.189 : Tensor = aten::_convolution(%256, %self.kp_detector.fg_encoder.layer3.1.conv1.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1/__module.kp_detector.fg_encoder.layer3.1.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.191 : Tensor = aten::batch_norm(%input.189, %self.kp_detector.fg_encoder.layer3.1.bn1.weight, %self.kp_detector.fg_encoder.layer3.1.bn1.bias, %self.kp_detector.fg_encoder.layer3.1.bn1.running_mean, %self.kp_detector.fg_encoder.layer3.1.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1/__module.kp_detector.fg_encoder.layer3.1.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %259 : Tensor = aten::relu(%input.191), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1/__module.kp_detector.fg_encoder.layer3.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0, Tactic: 0xa033e20ae9f412b2, (Unnamed Layer* 138) [Activation]_output (Half[1,256:8,16,16]) -> (Unnamed Layer* 141) [Activation]_output (Half[1,256:8,16,16])
Layer(CaskConvolution): %input.303 : Tensor = aten::_convolution(%335, %self.kp_detector.fg_encoder.layer3.1.conv1.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1/__module.kp_detector.fg_encoder.layer3.1.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.305 : Tensor = aten::batch_norm(%input.303, %self.kp_detector.fg_encoder.layer3.1.bn1.weight, %self.kp_detector.fg_encoder.layer3.1.bn1.bias, %self.kp_detector.fg_encoder.layer3.1.bn1.running_mean, %self.kp_detector.fg_encoder.layer3.1.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1/__module.kp_detector.fg_encoder.layer3.1.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %338 : Tensor = aten::relu(%input.305), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1/__module.kp_detector.fg_encoder.layer3.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0, Tactic: 0xa033e20ae9f412b2, (Unnamed Layer* 222) [Activation]_output (Half[1,256:8,16,16]) -> (Unnamed Layer* 225) [Activation]_output (Half[1,256:8,16,16])
Layer(CaskConvolution): %input.309 : Tensor = aten::_convolution(%338, %self.kp_detector.fg_encoder.layer3.1.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1/__module.kp_detector.fg_encoder.layer3.1.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.43 : Tensor = aten::batch_norm(%input.309, %self.kp_detector.fg_encoder.layer3.1.bn2.weight, %self.kp_detector.fg_encoder.layer3.1.bn2.bias, %self.kp_detector.fg_encoder.layer3.1.bn2.running_mean, %self.kp_detector.fg_encoder.layer3.1.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1/__module.kp_detector.fg_encoder.layer3.1.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %341 : Tensor = aten::add(%out.43, %335, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %342 : Tensor = aten::relu(%341), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1/__module.kp_detector.fg_encoder.layer3.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0, Tactic: 0xa033e20ae9f412b2, (Unnamed Layer* 225) [Activation]_output (Half[1,256:8,16,16]), (Unnamed Layer* 222) [Activation]_output (Half[1,256:8,16,16]) -> (Unnamed Layer* 229) [Activation]_output (Half[1,256:8,16,16])
Layer(CaskConvolution): %input.195 : Tensor = aten::_convolution(%259, %self.kp_detector.fg_encoder.layer3.1.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1/__module.kp_detector.fg_encoder.layer3.1.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.27 : Tensor = aten::batch_norm(%input.195, %self.kp_detector.fg_encoder.layer3.1.bn2.weight, %self.kp_detector.fg_encoder.layer3.1.bn2.bias, %self.kp_detector.fg_encoder.layer3.1.bn2.running_mean, %self.kp_detector.fg_encoder.layer3.1.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1/__module.kp_detector.fg_encoder.layer3.1.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %262 : Tensor = aten::add(%out.27, %256, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %263 : Tensor = aten::relu(%262), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer3/__module.kp_detector.fg_encoder.layer3.1/__module.kp_detector.fg_encoder.layer3.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0, Tactic: 0xa033e20ae9f412b2, (Unnamed Layer* 141) [Activation]_output (Half[1,256:8,16,16]), (Unnamed Layer* 138) [Activation]_output (Half[1,256:8,16,16]) -> (Unnamed Layer* 145) [Activation]_output (Half[1,256:8,16,16])
Layer(CaskConvolution): %input.201 : Tensor = aten::_convolution(%263, %self.kp_detector.fg_encoder.layer4.0.conv1.weight, %5, %25, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.203 : Tensor = aten::batch_norm(%input.201, %self.kp_detector.fg_encoder.layer4.0.bn1.weight, %self.kp_detector.fg_encoder.layer4.0.bn1.bias, %self.kp_detector.fg_encoder.layer4.0.bn1.running_mean, %self.kp_detector.fg_encoder.layer4.0.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %266 : Tensor = aten::relu(%input.203), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0, Tactic: 0xa033e20ae9f412b2, (Unnamed Layer* 145) [Activation]_output (Half[1,256:8,16,16]) -> (Unnamed Layer* 148) [Activation]_output (Half[1,512:8,8,8])
Layer(CaskConvolution): %input.315 : Tensor = aten::_convolution(%342, %self.kp_detector.fg_encoder.layer4.0.conv1.weight, %5, %25, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.317 : Tensor = aten::batch_norm(%input.315, %self.kp_detector.fg_encoder.layer4.0.bn1.weight, %self.kp_detector.fg_encoder.layer4.0.bn1.bias, %self.kp_detector.fg_encoder.layer4.0.bn1.running_mean, %self.kp_detector.fg_encoder.layer4.0.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %345 : Tensor = aten::relu(%input.317), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0, Tactic: 0xa033e20ae9f412b2, (Unnamed Layer* 229) [Activation]_output (Half[1,256:8,16,16]) -> (Unnamed Layer* 232) [Activation]_output (Half[1,512:8,8,8])
Layer(CaskConvolution): %input.321 : Tensor = aten::_convolution(%345, %self.kp_detector.fg_encoder.layer4.0.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.45 : Tensor = aten::batch_norm(%input.321, %self.kp_detector.fg_encoder.layer4.0.bn2.weight, %self.kp_detector.fg_encoder.layer4.0.bn2.bias, %self.kp_detector.fg_encoder.layer4.0.bn2.running_mean, %self.kp_detector.fg_encoder.layer4.0.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0, Tactic: 0xa033e20ae9f412b2, (Unnamed Layer* 232) [Activation]_output (Half[1,512:8,8,8]) -> (Unnamed Layer* 236) [Scale]_output (Half[1,512:8,8,8])
Layer(CaskConvolution): %input.207 : Tensor = aten::_convolution(%266, %self.kp_detector.fg_encoder.layer4.0.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.29 : Tensor = aten::batch_norm(%input.207, %self.kp_detector.fg_encoder.layer4.0.bn2.weight, %self.kp_detector.fg_encoder.layer4.0.bn2.bias, %self.kp_detector.fg_encoder.layer4.0.bn2.running_mean, %self.kp_detector.fg_encoder.layer4.0.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0, Tactic: 0xa033e20ae9f412b2, (Unnamed Layer* 148) [Activation]_output (Half[1,512:8,8,8]) -> (Unnamed Layer* 152) [Scale]_output (Half[1,512:8,8,8])
Layer(CaskConvolution): %input.323 : Tensor = aten::_convolution(%342, %self.kp_detector.fg_encoder.layer4.0.downsample.0.weight, %5, %25, %28, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.downsample/__module.kp_detector.fg_encoder.layer4.0.downsample.0 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %identity : Tensor = aten::batch_norm(%input.323, %self.kp_detector.fg_encoder.layer4.0.downsample.1.weight, %self.kp_detector.fg_encoder.layer4.0.bn2.bias, %self.kp_detector.fg_encoder.layer4.0.downsample.1.running_mean, %self.kp_detector.fg_encoder.layer4.0.downsample.1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.downsample/__module.kp_detector.fg_encoder.layer4.0.downsample.1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %350 : Tensor = aten::add(%out.45, %identity, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %351 : Tensor = aten::relu(%350), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0, Tactic: 0x2aa016c86360697f, (Unnamed Layer* 229) [Activation]_output (Half[1,256:8,16,16]), (Unnamed Layer* 236) [Scale]_output (Half[1,512:8,8,8]) -> (Unnamed Layer* 238) [Activation]_output (Half[1,512:8,8,8])
Layer(CaskConvolution): %input.209 : Tensor = aten::_convolution(%263, %self.kp_detector.fg_encoder.layer4.0.downsample.0.weight, %5, %25, %28, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.downsample/__module.kp_detector.fg_encoder.layer4.0.downsample.0 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %identity.11 : Tensor = aten::batch_norm(%input.209, %self.kp_detector.fg_encoder.layer4.0.downsample.1.weight, %self.kp_detector.fg_encoder.layer4.0.bn2.bias, %self.kp_detector.fg_encoder.layer4.0.downsample.1.running_mean, %self.kp_detector.fg_encoder.layer4.0.downsample.1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.downsample/__module.kp_detector.fg_encoder.layer4.0.downsample.1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %271 : Tensor = aten::add(%out.29, %identity.11, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %272 : Tensor = aten::relu(%271), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.0/__module.kp_detector.fg_encoder.layer4.0.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0, Tactic: 0x2aa016c86360697f, (Unnamed Layer* 145) [Activation]_output (Half[1,256:8,16,16]), (Unnamed Layer* 152) [Scale]_output (Half[1,512:8,8,8]) -> (Unnamed Layer* 154) [Activation]_output (Half[1,512:8,8,8])
Layer(CaskConvolution): %input.215 : Tensor = aten::_convolution(%272, %self.kp_detector.fg_encoder.layer4.1.conv1.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1/__module.kp_detector.fg_encoder.layer4.1.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.217 : Tensor = aten::batch_norm(%input.215, %self.kp_detector.fg_encoder.layer4.1.bn1.weight, %self.kp_detector.fg_encoder.layer4.1.bn1.bias, %self.kp_detector.fg_encoder.layer4.1.bn1.running_mean, %self.kp_detector.fg_encoder.layer4.1.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1/__module.kp_detector.fg_encoder.layer4.1.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %275 : Tensor = aten::relu(%input.217), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1/__module.kp_detector.fg_encoder.layer4.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0, Tactic: 0xa033e20ae9f412b2, (Unnamed Layer* 154) [Activation]_output (Half[1,512:8,8,8]) -> (Unnamed Layer* 157) [Activation]_output (Half[1,512:8,8,8])
Layer(CaskConvolution): %input.329 : Tensor = aten::_convolution(%351, %self.kp_detector.fg_encoder.layer4.1.conv1.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1/__module.kp_detector.fg_encoder.layer4.1.conv1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %input.331 : Tensor = aten::batch_norm(%input.329, %self.kp_detector.fg_encoder.layer4.1.bn1.weight, %self.kp_detector.fg_encoder.layer4.1.bn1.bias, %self.kp_detector.fg_encoder.layer4.1.bn1.running_mean, %self.kp_detector.fg_encoder.layer4.1.bn1.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1/__module.kp_detector.fg_encoder.layer4.1.bn1 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %354 : Tensor = aten::relu(%input.331), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1/__module.kp_detector.fg_encoder.layer4.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0, Tactic: 0xa033e20ae9f412b2, (Unnamed Layer* 238) [Activation]_output (Half[1,512:8,8,8]) -> (Unnamed Layer* 241) [Activation]_output (Half[1,512:8,8,8])
Layer(CaskConvolution): %input.335 : Tensor = aten::_convolution(%354, %self.kp_detector.fg_encoder.layer4.1.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1/__module.kp_detector.fg_encoder.layer4.1.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.47 : Tensor = aten::batch_norm(%input.335, %self.kp_detector.fg_encoder.layer4.1.bn2.weight, %self.kp_detector.fg_encoder.layer4.1.bn2.bias, %self.kp_detector.fg_encoder.layer4.1.bn2.running_mean, %self.kp_detector.fg_encoder.layer4.1.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1/__module.kp_detector.fg_encoder.layer4.1.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %357 : Tensor = aten::add(%out.47, %351, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %358 : Tensor = aten::relu(%357), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1/__module.kp_detector.fg_encoder.layer4.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0, Tactic: 0xa033e20ae9f412b2, (Unnamed Layer* 241) [Activation]_output (Half[1,512:8,8,8]), (Unnamed Layer* 238) [Activation]_output (Half[1,512:8,8,8]) -> (Unnamed Layer* 245) [Activation]_output (Half[1,512:8,8,8])
Layer(CaskPooling): %x.9 : Tensor = aten::adaptive_avg_pool2d(%358, %27), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.avgpool # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1213:0, Tactic: 0x56d7b61f084f251e, (Unnamed Layer* 245) [Activation]_output (Half[1,512:8,8,8]) -> (Unnamed Layer* 246) [Reduce]_output (Half[1,512:8,1,1])
Layer(CaskConvolution): %input.221 : Tensor = aten::_convolution(%275, %self.kp_detector.fg_encoder.layer4.1.conv2.weight, %5, %27, %27, %27, %4, %28, %365, %4, %4, %29, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1/__module.kp_detector.fg_encoder.layer4.1.conv2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/modules/conv.py:458:0 + %out.31 : Tensor = aten::batch_norm(%input.221, %self.kp_detector.fg_encoder.layer4.1.bn2.weight, %self.kp_detector.fg_encoder.layer4.1.bn2.bias, %self.kp_detector.fg_encoder.layer4.1.bn2.running_mean, %self.kp_detector.fg_encoder.layer4.1.bn2.running_var, %4, %35, %36, %29), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1/__module.kp_detector.fg_encoder.layer4.1.bn2 # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:2435:0 + %278 : Tensor = aten::add(%out.31, %272, %365), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1 # /home/joy/.venv/lib/python3.10/site-packages/torchvision/models/resnet.py:96:0 + %279 : Tensor = aten::relu(%278), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.layer4/__module.kp_detector.fg_encoder.layer4.1/__module.kp_detector.fg_encoder.layer4.1.relu # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1453:0, Tactic: 0xa033e20ae9f412b2, (Unnamed Layer* 157) [Activation]_output (Half[1,512:8,8,8]), (Unnamed Layer* 154) [Activation]_output (Half[1,512:8,8,8]) -> (Unnamed Layer* 161) [Activation]_output (Half[1,512:8,8,8])
Layer(CaskPooling): %x.7 : Tensor = aten::adaptive_avg_pool2d(%279, %27), scope: __module.kp_detector/__module.kp_detector.fg_encoder/__module.kp_detector.fg_encoder.avgpool # /home/joy/.venv/lib/python3.10/site-packages/torch/nn/functional.py:1213:0, Tactic: 0x56d7b61f084f251e, (Unnamed Layer* 161) [Activation]_output (Half[1,512:8,8,8]) -> (Unnamed Layer* 162) [Reduce]_output (Half[1,512:8,1,1])
Layer(NoOp): Reformatting CopyNode for Input Tensor 0 to %362 : Tensor = aten::matmul(%input.341, %361) + [Freeze Tensor %363 : Tensor = trt::const(%self.kp_detector.fg_encoder.fc.bias) ] + (Unnamed Layer* 252) [Shuffle] + unsqueeze_node_after_[Freeze Tensor %363 : Tensor = trt::const(%self.kp_detector.fg_encoder.fc.bias) ] + (Unnamed Layer* 252) [Shuffle]_(Unnamed Layer* 252) [Shuffle]_output + %364 : Tensor = aten::add(%363, %362, %365) + PWN(%fg_kp.15 : Tensor = aten::sigmoid(%364), scope: __module.kp_detector # /home/joy/remote/stable-diffusion-backend/personalize/thin_plate_spline_motion_model/tpsnn.py:43:0), Tactic: 0x0000000000000000, (Unnamed Layer* 246) [Reduce]_output (Half[1,512:8,1,1]) -> Reformatted Input Tensor 0 to %362 : Tensor = aten::matmul(%input.341, %361) + [Freeze Tensor %363 : Tensor = trt::const(%self.kp_detector.fg_encoder.fc.bias) ] + (Unnamed Layer* 252) [Shuffle] + unsqueeze_node_after_[Freeze Tensor %363 : Tensor = trt::const(%self.kp_detector.fg_encoder.fc.bias) ] + (Unnamed Layer* 252) [Shuffle]_(Unnamed Layer* 252) [Shuffle]_output + %364 : Tensor = aten::add(%363, %362, %365) + PWN(%fg_kp.15 : Tensor = aten::sigmoid(%364), scope: __module.kp_detector # /home/joy/remote/stable-diffusion-backend/personalize/thin_plate_spline_motion_model/tpsnn.py:43:0) (Half[1,512,1,1])
Layer(CublasConvolution): %362 : Tensor = aten::matmul(%input.341, %361) + [Freeze Tensor %363 : Tensor = trt::const(%self.kp_detector.fg_encoder.fc.bias) ] + (Unnamed Layer* 252) [Shuffle] + unsqueeze_node_after_[Freeze Tensor %363 : Tensor = trt::const(%self.kp_detector.fg_encoder.fc.bias) ] + (Unnamed Layer* 252) [Shuffle]_(Unnamed Layer* 252) [Shuffle]_output + %364 : Tensor = aten::add(%363, %362, %365) + PWN(%fg_kp.15 : Tensor = aten::sigmoid(%364), scope: __module.kp_detector # /home/joy/remote/stable-diffusion-backend/personalize/thin_plate_spline_motion_model/tpsnn.py:43:0), Tactic: 0x0000000000000002, Reformatted Input Tensor 0 to %362 : Tensor = aten::matmul(%input.341, %361) + [Freeze Tensor %363 : Tensor = trt::const(%self.kp_detector.fg_encoder.fc.bias) ] + (Unnamed Layer* 252) [Shuffle] + unsqueeze_node_after_[Freeze Tensor %363 : Tensor = trt::const(%self.kp_detector.fg_encoder.fc.bias) ] + (Unnamed Layer* 252) [Shuffle]_(Unnamed Layer* 252) [Shuffle]_output + %364 : Tensor = aten::add(%363, %362, %365) + PWN(%fg_kp.15 : Tensor = aten::sigmoid(%364), scope: __module.kp_detector # /home/joy/remote/stable-diffusion-backend/personalize/thin_plate_spline_motion_model/tpsnn.py:43:0) (Half[1,512,1,1]) -> %fg_kp.15 : Tensor = aten::sigmoid(%364), scope: __module.kp_detector # /home/joy/remote/stable-diffusion-backend/personalize/thin_plate_spline_motion_model/tpsnn.py:43:0_out_tensor (Half[1,100,1,1])
Layer(NoOp): Reformatting CopyNode for Input Tensor 0 to %283 : Tensor = aten::matmul(%input.227, %282) + [Freeze Tensor %284 : Tensor = trt::const(%self.kp_detector.fg_encoder.fc.bias) ] + (Unnamed Layer* 168) [Shuffle] + unsqueeze_node_after_[Freeze Tensor %284 : Tensor = trt::const(%self.kp_detector.fg_encoder.fc.bias) ] + (Unnamed Layer* 168) [Shuffle]_(Unnamed Layer* 168) [Shuffle]_output + %285 : Tensor = aten::add(%284, %283, %365) + PWN(%fg_kp.9 : Tensor = aten::sigmoid(%285), scope: __module.kp_detector # /home/joy/remote/stable-diffusion-backend/personalize/thin_plate_spline_motion_model/tpsnn.py:43:0), Tactic: 0x0000000000000000, (Unnamed Layer* 162) [Reduce]_output (Half[1,512:8,1,1]) -> Reformatted Input Tensor 0 to %283 : Tensor = aten::matmul(%input.227, %282) + [Freeze Tensor %284 : Tensor = trt::const(%self.kp_detector.fg_encoder.fc.bias) ] + (Unnamed Layer* 168) [Shuffle] + unsqueeze_node_after_[Freeze Tensor %284 : Tensor = trt::const(%self.kp_detector.fg_encoder.fc.bias) ] + (Unnamed Layer* 168) [Shuffle]_(Unnamed Layer* 168) [Shuffle]_output + %285 : Tensor = aten::add(%284, %283, %365) + PWN(%fg_kp.9 : Tensor = aten::sigmoid(%285), scope: __module.kp_detector # /home/joy/remote/stable-diffusion-backend/personalize/thin_plate_spline_motion_model/tpsnn.py:43:0) (Half[1,512,1,1])
Layer(CublasConvolution): %283 : Tensor = aten::matmul(%input.227, %282) + [Freeze Tensor %284 : Tensor = trt::const(%self.kp_detector.fg_encoder.fc.bias) ] + (Unnamed Layer* 168) [Shuffle] + unsqueeze_node_after_[Freeze Tensor %284 : Tensor = trt::const(%self.kp_detector.fg_encoder.fc.bias) ] + (Unnamed Layer* 168) [Shuffle]_(Unnamed Layer* 168) [Shuffle]_output + %285 : Tensor = aten::add(%284, %283, %365) + PWN(%fg_kp.9 : Tensor = aten::sigmoid(%285), scope: __module.kp_detector # /home/joy/remote/stable-diffusion-backend/personalize/thin_plate_spline_motion_model/tpsnn.py:43:0), Tactic: 0x0000000000000002, Reformatted Input Tensor 0 to %283 : Tensor = aten::matmul(%input.227, %282) + [Freeze Tensor %284 : Tensor = trt::const(%self.kp_detector.fg_encoder.fc.bias) ] + (Unnamed Layer* 168) [Shuffle] + unsqueeze_node_after_[Freeze Tensor %284 : Tensor = trt::const(%self.kp_detector.fg_encoder.fc.bias) ] + (Unnamed Layer* 168) [Shuffle]_(Unnamed Layer* 168) [Shuffle]_output + %285 : Tensor = aten::add(%284, %283, %365) + PWN(%fg_kp.9 : Tensor = aten::sigmoid(%285), scope: __module.kp_detector # /home/joy/remote/stable-diffusion-backend/personalize/thin_plate_spline_motion_model/tpsnn.py:43:0) (Half[1,512,1,1]) -> %fg_kp.9 : Tensor = aten::sigmoid(%285), scope: __module.kp_detector # /home/joy/remote/stable-diffusion-backend/personalize/thin_plate_spline_motion_model/tpsnn.py:43:0_out_tensor (Half[1,100,1,1])
Layer(Reformat): Reformatting CopyNode for Input Tensor 0 to %368 : Tensor = aten::mul(%fg_kp.15, %206), scope: __module.kp_detector # /home/joy/remote/stable-diffusion-backend/personalize/thin_plate_spline_motion_model/tpsnn.py:43:0, Tactic: 0x00000000000003e8, %fg_kp.15 : Tensor = aten::sigmoid(%364), scope: __module.kp_detector # /home/joy/remote/stable-diffusion-backend/personalize/thin_plate_spline_motion_model/tpsnn.py:43:0_out_tensor (Half[1,100,1,1]) -> Reformatted Input Tensor 0 to %368 : Tensor = aten::mul(%fg_kp.15, %206), scope: __module.kp_detector # /home/joy/remote/stable-diffusion-backend/personalize/thin_plate_spline_motion_model/tpsnn.py:43:0 (Float[1,100,1,1])
Layer(ElementWise): %368 : Tensor = aten::mul(%fg_kp.15, %206), scope: __module.kp_detector # /home/joy/remote/stable-diffusion-backend/personalize/thin_plate_spline_motion_model/tpsnn.py:43:0, Tactic: 0x0000000000000001, Reformatted Input Tensor 0 to %368 : Tensor = aten::mul(%fg_kp.15, %206), scope: __module.kp_detector # /home/joy/remote/stable-diffusion-backend/personalize/thin_plate_spline_motion_model/tpsnn.py:43:0 (Float[1,100,1,1]), unsqueeze_tensor_after_{ForeignNode[(Unnamed Layer* 1) [Shuffle]...%input.115 : Tensor = aten::select(%19, %21, %16) # /home/joy/remote/stable-diffusion-backend/personalize/thin_plate_spline_motion_model/tpsnn.py:529:0]}_(Unnamed Layer* 257) [Shuffle]_output_out_tensor (Float[1,1,1,1]) -> %368 : Tensor = aten::mul(%fg_kp.15, %206), scope: __module.kp_detector # /home/joy/remote/stable-diffusion-backend/personalize/thin_plate_spline_motion_model/tpsnn.py:43:0_out_tensor (Float[1,100,1,1])
Layer(NoOp): copied_squeeze_after_%368 : Tensor = aten::mul(%fg_kp.15, %206), scope: __module.kp_detector # /home/joy/remote/stable-diffusion-backend/personalize/thin_plate_spline_motion_model/tpsnn.py:43:0, Tactic: 0x0000000000000000, %368 : Tensor = aten::mul(%fg_kp.15, %206), scope: __module.kp_detector # /home/joy/remote/stable-diffusion-backend/personalize/thin_plate_spline_motion_model/tpsnn.py:43:0_out_tensor (Float[1,100,1,1]) -> (Unnamed Layer* 258) [ElementWise]_output (Float[1,100])
Layer(ElementWise): %fg_kp : Tensor = aten::sub(%368, %208, %365), scope: __module.kp_detector # /home/joy/remote/stable-diffusion-backend/personalize/thin_plate_spline_motion_model/tpsnn.py:43:0, Tactic: 0x0000000000000001, (Unnamed Layer* 258) [ElementWise]_output (Float[1,100]), (Unnamed Layer* 261) [Shuffle]_output (Float[1,1]) -> (Unnamed Layer* 262) [ElementWise]_output (Float[1,100])
Layer(NoOp): %371 : Tensor = aten::reshape(%fg_kp, %370), Tactic: 0x0000000000000000, (Unnamed Layer* 262) [ElementWise]_output (Float[1,100]) -> (Unnamed Layer* 263) [Shuffle]_output (Float[1,50,2])
Layer(Reformat): Reformatting CopyNode for Input Tensor 0 to %289 : Tensor = aten::mul(%fg_kp.9, %206), scope: __module.kp_detector # /home/joy/remote/stable-diffusion-backend/personalize/thin_plate_spline_motion_model/tpsnn.py:43:0, Tactic: 0x00000000000003e8, %fg_kp.9 : Tensor = aten::sigmoid(%285), scope: __module.kp_detector # /home/joy/remote/stable-diffusion-backend/personalize/thin_plate_spline_motion_model/tpsnn.py:43:0_out_tensor (Half[1,100,1,1]) -> Reformatted Input Tensor 0 to %289 : Tensor = aten::mul(%fg_kp.9, %206), scope: __module.kp_detector # /home/joy/remote/stable-diffusion-backend/personalize/thin_plate_spline_motion_model/tpsnn.py:43:0 (Float[1,100,1,1])
Layer(ElementWise): %289 : Tensor = aten::mul(%fg_kp.9, %206), scope: __module.kp_detector # /home/joy/remote/stable-diffusion-backend/personalize/thin_plate_spline_motion_model/tpsnn.py:43:0, Tactic: 0x0000000000000001, Reformatted Input Tensor 0 to %289 : Tensor = aten::mul(%fg_kp.9, %206), scope: __module.kp_detector # /home/joy/remote/stable-diffusion-backend/personalize/thin_plate_spline_motion_model/tpsnn.py:43:0 (Float[1,100,1,1]), unsqueeze_tensor_after_{Foreign