tvm-commits mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From GitBox <...@apache.org>
Subject [GitHub] [incubator-tvm] qinxianyuzi opened a new issue #5218: TVMError: Check failed: ObjectTypeChecker: :Check(ptr): Expect relay.Expr but get IRModule
Date Thu, 02 Apr 2020 08:25:49 GMT
qinxianyuzi opened a new issue #5218: TVMError: Check failed: ObjectTypeChecker: :Check(ptr):
Expect relay.Expr but get IRModule
URL: https://github.com/apache/incubator-tvm/issues/5218
 
 
   When I use this code for evaluating, An TVMError has occurred.
   `with relay.build_config(opt_level=0):
       intrp = relay.build_module.create_executor('graph', sym, tvm.cpu(0), target)
   dtype = 'float32'
   func = intrp.evaluate(sym)`
   
   TVMError: Check failed: ObjectTypeChecker: :Check(ptr): Expect relay.Expr but get IRModule
   
   
   The onnx graph:
   `
   graph(%input.1 : Float(1, 3, 224, 224),
         %conv1.weight : Float(32, 3, 3, 3),
         %bn1.weight : Float(32),
         %bn1.bias : Float(32),
         %bn1.running_mean : Float(32),
         %bn1.running_var : Float(32),
         %bn1.num_batches_tracked : Long(),
         %layer1.0.conv1.weight : Float(32, 32, 3, 3),
         %layer1.0.bn2.weight : Float(32),
         %layer1.0.bn2.bias : Float(32),
         %layer1.0.bn2.running_mean : Float(32),
         %layer1.0.bn2.running_var : Float(32),
         %layer1.0.bn2.num_batches_tracked : Long(),
         %layer1.0.conv2.weight : Float(32, 32, 3, 3),
         %layer1.0.bn3.weight : Float(32),
         %layer1.0.bn3.bias : Float(32),
         %layer1.0.bn3.running_mean : Float(32),
         %layer1.0.bn3.running_var : Float(32),
         %layer1.0.bn3.num_batches_tracked : Long(),
         %layer1.0.downsample.0.weight : Float(32, 32, 1, 1),
         %layer1.0.downsample.1.weight : Float(32),
         %layer1.0.downsample.1.bias : Float(32),
         %layer1.0.downsample.1.running_mean : Float(32),
         %layer1.0.downsample.1.running_var : Float(32),
         %layer1.0.downsample.1.num_batches_tracked : Long(),
         %layer1.1.conv1.weight : Float(32, 32, 3, 3),
         %layer1.1.bn2.weight : Float(32),
         %layer1.1.bn2.bias : Float(32),
         %layer1.1.bn2.running_mean : Float(32),
         %layer1.1.bn2.running_var : Float(32),
         %layer1.1.bn2.num_batches_tracked : Long(),
         %layer1.1.conv2.weight : Float(32, 32, 3, 3),
         %layer1.1.bn3.weight : Float(32),
         %layer1.1.bn3.bias : Float(32),
         %layer1.1.bn3.running_mean : Float(32),
         %layer1.1.bn3.running_var : Float(32),
         %layer1.1.bn3.num_batches_tracked : Long(),
         %layer2.0.conv1.weight : Float(64, 32, 3, 3),
         %layer2.0.bn2.weight : Float(64),
         %layer2.0.bn2.bias : Float(64),
         %layer2.0.bn2.running_mean : Float(64),
         %layer2.0.bn2.running_var : Float(64),
         %layer2.0.bn2.num_batches_tracked : Long(),
         %layer2.0.conv2.weight : Float(64, 64, 3, 3),
         %layer2.0.bn3.weight : Float(64),
         %layer2.0.bn3.bias : Float(64),
         %layer2.0.bn3.running_mean : Float(64),
         %layer2.0.bn3.running_var : Float(64),
         %layer2.0.bn3.num_batches_tracked : Long(),
         %layer2.0.downsample.0.weight : Float(64, 32, 1, 1),
         %layer2.0.downsample.1.weight : Float(64),
         %layer2.0.downsample.1.bias : Float(64),
         %layer2.0.downsample.1.running_mean : Float(64),
         %layer2.0.downsample.1.running_var : Float(64),
         %layer2.0.downsample.1.num_batches_tracked : Long(),
         %layer2.1.conv1.weight : Float(64, 64, 3, 3),
         %layer2.1.bn2.weight : Float(64),
         %layer2.1.bn2.bias : Float(64),
         %layer2.1.bn2.running_mean : Float(64),
         %layer2.1.bn2.running_var : Float(64),
         %layer2.1.bn2.num_batches_tracked : Long(),
         %layer2.1.conv2.weight : Float(64, 64, 3, 3),
         %layer2.1.bn3.weight : Float(64),
         %layer2.1.bn3.bias : Float(64),
         %layer2.1.bn3.running_mean : Float(64),
         %layer2.1.bn3.running_var : Float(64),
         %layer2.1.bn3.num_batches_tracked : Long(),
         %layer3.0.conv1.weight : Float(128, 64, 3, 3),
         %layer3.0.bn2.weight : Float(128),
         %layer3.0.bn2.bias : Float(128),
         %layer3.0.bn2.running_mean : Float(128),
         %layer3.0.bn2.running_var : Float(128),
         %layer3.0.bn2.num_batches_tracked : Long(),
         %layer3.0.conv2.weight : Float(128, 128, 3, 3),
         %layer3.0.bn3.weight : Float(128),
         %layer3.0.bn3.bias : Float(128),
         %layer3.0.bn3.running_mean : Float(128),
         %layer3.0.bn3.running_var : Float(128),
         %layer3.0.bn3.num_batches_tracked : Long(),
         %layer3.0.downsample.0.weight : Float(128, 64, 1, 1),
         %layer3.0.downsample.1.weight : Float(128),
         %layer3.0.downsample.1.bias : Float(128),
         %layer3.0.downsample.1.running_mean : Float(128),
         %layer3.0.downsample.1.running_var : Float(128),
         %layer3.0.downsample.1.num_batches_tracked : Long(),
         %layer3.1.conv1.weight : Float(128, 128, 3, 3),
         %layer3.1.bn2.weight : Float(128),
         %layer3.1.bn2.bias : Float(128),
         %layer3.1.bn2.running_mean : Float(128),
         %layer3.1.bn2.running_var : Float(128),
         %layer3.1.bn2.num_batches_tracked : Long(),
         %layer3.1.conv2.weight : Float(128, 128, 3, 3),
         %layer3.1.bn3.weight : Float(128),
         %layer3.1.bn3.bias : Float(128),
         %layer3.1.bn3.running_mean : Float(128),
         %layer3.1.bn3.running_var : Float(128),
         %layer3.1.bn3.num_batches_tracked : Long(),
         %layer4.0.conv1.weight : Float(384, 128, 3, 3),
         %layer4.0.bn2.weight : Float(384),
         %layer4.0.bn2.bias : Float(384),
         %layer4.0.bn2.running_mean : Float(384),
         %layer4.0.bn2.running_var : Float(384),
         %layer4.0.bn2.num_batches_tracked : Long(),
         %layer4.0.conv2.weight : Float(384, 384, 3, 3),
         %layer4.0.bn3.weight : Float(384),
         %layer4.0.bn3.bias : Float(384),
         %layer4.0.bn3.running_mean : Float(384),
         %layer4.0.bn3.running_var : Float(384),
         %layer4.0.bn3.num_batches_tracked : Long(),
         %layer4.0.downsample.0.weight : Float(384, 128, 1, 1),
         %layer4.0.downsample.1.weight : Float(384),
         %layer4.0.downsample.1.bias : Float(384),
         %layer4.0.downsample.1.running_mean : Float(384),
         %layer4.0.downsample.1.running_var : Float(384),
         %layer4.0.downsample.1.num_batches_tracked : Long(),
         %layer4.1.conv1.weight : Float(384, 384, 3, 3),
         %layer4.1.bn2.weight : Float(384),
         %layer4.1.bn2.bias : Float(384),
         %layer4.1.bn2.running_mean : Float(384),
         %layer4.1.bn2.running_var : Float(384),
         %layer4.1.bn2.num_batches_tracked : Long(),
         %layer4.1.conv2.weight : Float(384, 384, 3, 3),
         %layer4.1.bn3.weight : Float(384),
         %layer4.1.bn3.bias : Float(384),
         %layer4.1.bn3.running_mean : Float(384),
         %layer4.1.bn3.running_var : Float(384),
         %layer4.1.bn3.num_batches_tracked : Long(),
         %classifier.weight : Float(8, 384),
         %classifier.bias : Float(8)):
     %129 : Float(1, 32, 112, 112) = onnx::Conv[dilations=[1, 1], group=1, kernel_shape=[3,
3], pads=[1, 1, 1, 1], strides=[2, 2]](%input.1, %conv1.weight), scope: LResNet2/Conv2d[conv1]
# /home/huangry/anaconda3/lib/python3.7/site-packages/torch/nn/modules/conv.py:342:0
     %130 : Float(1, 32, 112, 112) = onnx::BatchNormalization[epsilon=1e-05, momentum=0.9](%129,
%bn1.weight, %bn1.bias, %bn1.running_mean, %bn1.running_var), scope: LResNet2/BatchNorm2d[bn1]
# /home/huangry/anaconda3/lib/python3.7/site-packages/torch/nn/functional.py:1670:0
     %131 : Float(1, 32, 112, 112) = onnx::Relu(%130), scope: LResNet2/ReLU[ReLU1] # /home/huangry/anaconda3/lib/python3.7/site-packages/torch/nn/functional.py:912:0
     %132 : Float(1, 32, 112, 112) = onnx::Conv[dilations=[1, 1], group=1, kernel_shape=[3,
3], pads=[1, 1, 1, 1], strides=[1, 1]](%131, %layer1.0.conv1.weight), scope: LResNet2/Sequential[layer1]/BlockIR2[0]/Conv2d[conv1]
# /home/huangry/anaconda3/lib/python3.7/site-packages/torch/nn/modules/conv.py:342:0
     %133 : Float(1, 32, 112, 112) = onnx::BatchNormalization[epsilon=1e-05, momentum=0.9](%132,
%layer1.0.bn2.weight, %layer1.0.bn2.bias, %layer1.0.bn2.running_mean, %layer1.0.bn2.running_var),
scope: LResNet2/Sequential[layer1]/BlockIR2[0]/BatchNorm2d[bn2] # /home/huangry/anaconda3/lib/python3.7/site-packages/torch/nn/functional.py:1670:0
     %134 : Float(1, 32, 112, 112) = onnx::Relu(%133), scope: LResNet2/Sequential[layer1]/BlockIR2[0]/ReLU[ReLU1]
# /home/huangry/anaconda3/lib/python3.7/site-packages/torch/nn/functional.py:912:0
     %135 : Float(1, 32, 56, 56) = onnx::Conv[dilations=[1, 1], group=1, kernel_shape=[3,
3], pads=[1, 1, 1, 1], strides=[2, 2]](%134, %layer1.0.conv2.weight), scope: LResNet2/Sequential[layer1]/BlockIR2[0]/Conv2d[conv2]
# /home/huangry/anaconda3/lib/python3.7/site-packages/torch/nn/modules/conv.py:342:0
     %136 : Float(1, 32, 56, 56) = onnx::BatchNormalization[epsilon=1e-05, momentum=0.9](%135,
%layer1.0.bn3.weight, %layer1.0.bn3.bias, %layer1.0.bn3.running_mean, %layer1.0.bn3.running_var),
scope: LResNet2/Sequential[layer1]/BlockIR2[0]/BatchNorm2d[bn3] # /home/huangry/anaconda3/lib/python3.7/site-packages/torch/nn/functional.py:1670:0
     %137 : Float(1, 32, 56, 56) = onnx::Conv[dilations=[1, 1], group=1, kernel_shape=[1,
1], pads=[0, 0, 0, 0], strides=[2, 2]](%131, %layer1.0.downsample.0.weight), scope: LResNet2/Sequential[layer1]/BlockIR2[0]/Sequential[downsample]/Conv2d[0]
# /home/huangry/anaconda3/lib/python3.7/site-packages/torch/nn/modules/conv.py:342:0
     %138 : Float(1, 32, 56, 56) = onnx::BatchNormalization[epsilon=1e-05, momentum=0.9](%137,
%layer1.0.downsample.1.weight, %layer1.0.downsample.1.bias, %layer1.0.downsample.1.running_mean,
%layer1.0.downsample.1.running_var), scope: LResNet2/Sequential[layer1]/BlockIR2[0]/Sequential[downsample]/BatchNorm2d[1]
# /home/huangry/anaconda3/lib/python3.7/site-packages/torch/nn/functional.py:1670:0
     %139 : Float(1, 32, 56, 56) = onnx::Add(%136, %138), scope: LResNet2/Sequential[layer1]/BlockIR2[0]
# /home/huangry/program/LResNet18E.py:44:0
     %140 : Float(1, 32, 56, 56) = onnx::Conv[dilations=[1, 1], group=1, kernel_shape=[3,
3], pads=[1, 1, 1, 1], strides=[1, 1]](%139, %layer1.1.conv1.weight), scope: LResNet2/Sequential[layer1]/BlockIR2[1]/Conv2d[conv1]
# /home/huangry/anaconda3/lib/python3.7/site-packages/torch/nn/modules/conv.py:342:0
     %141 : Float(1, 32, 56, 56) = onnx::BatchNormalization[epsilon=1e-05, momentum=0.9](%140,
%layer1.1.bn2.weight, %layer1.1.bn2.bias, %layer1.1.bn2.running_mean, %layer1.1.bn2.running_var),
scope: LResNet2/Sequential[layer1]/BlockIR2[1]/BatchNorm2d[bn2] # /home/huangry/anaconda3/lib/python3.7/site-packages/torch/nn/functional.py:1670:0
     %142 : Float(1, 32, 56, 56) = onnx::Relu(%141), scope: LResNet2/Sequential[layer1]/BlockIR2[1]/ReLU[ReLU1]
# /home/huangry/anaconda3/lib/python3.7/site-packages/torch/nn/functional.py:912:0
     %143 : Float(1, 32, 56, 56) = onnx::Conv[dilations=[1, 1], group=1, kernel_shape=[3,
3], pads=[1, 1, 1, 1], strides=[1, 1]](%142, %layer1.1.conv2.weight), scope: LResNet2/Sequential[layer1]/BlockIR2[1]/Conv2d[conv2]
# /home/huangry/anaconda3/lib/python3.7/site-packages/torch/nn/modules/conv.py:342:0
     %144 : Float(1, 32, 56, 56) = onnx::BatchNormalization[epsilon=1e-05, momentum=0.9](%143,
%layer1.1.bn3.weight, %layer1.1.bn3.bias, %layer1.1.bn3.running_mean, %layer1.1.bn3.running_var),
scope: LResNet2/Sequential[layer1]/BlockIR2[1]/BatchNorm2d[bn3] # /home/huangry/anaconda3/lib/python3.7/site-packages/torch/nn/functional.py:1670:0
     %145 : Float(1, 32, 56, 56) = onnx::Add(%144, %139), scope: LResNet2/Sequential[layer1]/BlockIR2[1]
# /home/huangry/program/LResNet18E.py:44:0
     %146 : Float(1, 64, 56, 56) = onnx::Conv[dilations=[1, 1], group=1, kernel_shape=[3,
3], pads=[1, 1, 1, 1], strides=[1, 1]](%145, %layer2.0.conv1.weight), scope: LResNet2/Sequential[layer2]/BlockIR2[0]/Conv2d[conv1]
# /home/huangry/anaconda3/lib/python3.7/site-packages/torch/nn/modules/conv.py:342:0
     %147 : Float(1, 64, 56, 56) = onnx::BatchNormalization[epsilon=1e-05, momentum=0.9](%146,
%layer2.0.bn2.weight, %layer2.0.bn2.bias, %layer2.0.bn2.running_mean, %layer2.0.bn2.running_var),
scope: LResNet2/Sequential[layer2]/BlockIR2[0]/BatchNorm2d[bn2] # /home/huangry/anaconda3/lib/python3.7/site-packages/torch/nn/functional.py:1670:0
     %148 : Float(1, 64, 56, 56) = onnx::Relu(%147), scope: LResNet2/Sequential[layer2]/BlockIR2[0]/ReLU[ReLU1]
# /home/huangry/anaconda3/lib/python3.7/site-packages/torch/nn/functional.py:912:0
     %149 : Float(1, 64, 28, 28) = onnx::Conv[dilations=[1, 1], group=1, kernel_shape=[3,
3], pads=[1, 1, 1, 1], strides=[2, 2]](%148, %layer2.0.conv2.weight), scope: LResNet2/Sequential[layer2]/BlockIR2[0]/Conv2d[conv2]
# /home/huangry/anaconda3/lib/python3.7/site-packages/torch/nn/modules/conv.py:342:0
     %150 : Float(1, 64, 28, 28) = onnx::BatchNormalization[epsilon=1e-05, momentum=0.9](%149,
%layer2.0.bn3.weight, %layer2.0.bn3.bias, %layer2.0.bn3.running_mean, %layer2.0.bn3.running_var),
scope: LResNet2/Sequential[layer2]/BlockIR2[0]/BatchNorm2d[bn3] # /home/huangry/anaconda3/lib/python3.7/site-packages/torch/nn/functional.py:1670:0
     %151 : Float(1, 64, 28, 28) = onnx::Conv[dilations=[1, 1], group=1, kernel_shape=[1,
1], pads=[0, 0, 0, 0], strides=[2, 2]](%145, %layer2.0.downsample.0.weight), scope: LResNet2/Sequential[layer2]/BlockIR2[0]/Sequential[downsample]/Conv2d[0]
# /home/huangry/anaconda3/lib/python3.7/site-packages/torch/nn/modules/conv.py:342:0
     %152 : Float(1, 64, 28, 28) = onnx::BatchNormalization[epsilon=1e-05, momentum=0.9](%151,
%layer2.0.downsample.1.weight, %layer2.0.downsample.1.bias, %layer2.0.downsample.1.running_mean,
%layer2.0.downsample.1.running_var), scope: LResNet2/Sequential[layer2]/BlockIR2[0]/Sequential[downsample]/BatchNorm2d[1]
# /home/huangry/anaconda3/lib/python3.7/site-packages/torch/nn/functional.py:1670:0
     %153 : Float(1, 64, 28, 28) = onnx::Add(%150, %152), scope: LResNet2/Sequential[layer2]/BlockIR2[0]
# /home/huangry/program/LResNet18E.py:44:0
     %154 : Float(1, 64, 28, 28) = onnx::Conv[dilations=[1, 1], group=1, kernel_shape=[3,
3], pads=[1, 1, 1, 1], strides=[1, 1]](%153, %layer2.1.conv1.weight), scope: LResNet2/Sequential[layer2]/BlockIR2[1]/Conv2d[conv1]
# /home/huangry/anaconda3/lib/python3.7/site-packages/torch/nn/modules/conv.py:342:0
     %155 : Float(1, 64, 28, 28) = onnx::BatchNormalization[epsilon=1e-05, momentum=0.9](%154,
%layer2.1.bn2.weight, %layer2.1.bn2.bias, %layer2.1.bn2.running_mean, %layer2.1.bn2.running_var),
scope: LResNet2/Sequential[layer2]/BlockIR2[1]/BatchNorm2d[bn2] # /home/huangry/anaconda3/lib/python3.7/site-packages/torch/nn/functional.py:1670:0
     %156 : Float(1, 64, 28, 28) = onnx::Relu(%155), scope: LResNet2/Sequential[layer2]/BlockIR2[1]/ReLU[ReLU1]
# /home/huangry/anaconda3/lib/python3.7/site-packages/torch/nn/functional.py:912:0
     %157 : Float(1, 64, 28, 28) = onnx::Conv[dilations=[1, 1], group=1, kernel_shape=[3,
3], pads=[1, 1, 1, 1], strides=[1, 1]](%156, %layer2.1.conv2.weight), scope: LResNet2/Sequential[layer2]/BlockIR2[1]/Conv2d[conv2]
# /home/huangry/anaconda3/lib/python3.7/site-packages/torch/nn/modules/conv.py:342:0
     %158 : Float(1, 64, 28, 28) = onnx::BatchNormalization[epsilon=1e-05, momentum=0.9](%157,
%layer2.1.bn3.weight, %layer2.1.bn3.bias, %layer2.1.bn3.running_mean, %layer2.1.bn3.running_var),
scope: LResNet2/Sequential[layer2]/BlockIR2[1]/BatchNorm2d[bn3] # /home/huangry/anaconda3/lib/python3.7/site-packages/torch/nn/functional.py:1670:0
     %159 : Float(1, 64, 28, 28) = onnx::Add(%158, %153), scope: LResNet2/Sequential[layer2]/BlockIR2[1]
# /home/huangry/program/LResNet18E.py:44:0
     %160 : Float(1, 128, 28, 28) = onnx::Conv[dilations=[1, 1], group=1, kernel_shape=[3,
3], pads=[1, 1, 1, 1], strides=[1, 1]](%159, %layer3.0.conv1.weight), scope: LResNet2/Sequential[layer3]/BlockIR2[0]/Conv2d[conv1]
# /home/huangry/anaconda3/lib/python3.7/site-packages/torch/nn/modules/conv.py:342:0
     %161 : Float(1, 128, 28, 28) = onnx::BatchNormalization[epsilon=1e-05, momentum=0.9](%160,
%layer3.0.bn2.weight, %layer3.0.bn2.bias, %layer3.0.bn2.running_mean, %layer3.0.bn2.running_var),
scope: LResNet2/Sequential[layer3]/BlockIR2[0]/BatchNorm2d[bn2] # /home/huangry/anaconda3/lib/python3.7/site-packages/torch/nn/functional.py:1670:0
     %162 : Float(1, 128, 28, 28) = onnx::Relu(%161), scope: LResNet2/Sequential[layer3]/BlockIR2[0]/ReLU[ReLU1]
# /home/huangry/anaconda3/lib/python3.7/site-packages/torch/nn/functional.py:912:0
     %163 : Float(1, 128, 14, 14) = onnx::Conv[dilations=[1, 1], group=1, kernel_shape=[3,
3], pads=[1, 1, 1, 1], strides=[2, 2]](%162, %layer3.0.conv2.weight), scope: LResNet2/Sequential[layer3]/BlockIR2[0]/Conv2d[conv2]
# /home/huangry/anaconda3/lib/python3.7/site-packages/torch/nn/modules/conv.py:342:0
     %164 : Float(1, 128, 14, 14) = onnx::BatchNormalization[epsilon=1e-05, momentum=0.9](%163,
%layer3.0.bn3.weight, %layer3.0.bn3.bias, %layer3.0.bn3.running_mean, %layer3.0.bn3.running_var),
scope: LResNet2/Sequential[layer3]/BlockIR2[0]/BatchNorm2d[bn3] # /home/huangry/anaconda3/lib/python3.7/site-packages/torch/nn/functional.py:1670:0
     %165 : Float(1, 128, 14, 14) = onnx::Conv[dilations=[1, 1], group=1, kernel_shape=[1,
1], pads=[0, 0, 0, 0], strides=[2, 2]](%159, %layer3.0.downsample.0.weight), scope: LResNet2/Sequential[layer3]/BlockIR2[0]/Sequential[downsample]/Conv2d[0]
# /home/huangry/anaconda3/lib/python3.7/site-packages/torch/nn/modules/conv.py:342:0
     %166 : Float(1, 128, 14, 14) = onnx::BatchNormalization[epsilon=1e-05, momentum=0.9](%165,
%layer3.0.downsample.1.weight, %layer3.0.downsample.1.bias, %layer3.0.downsample.1.running_mean,
%layer3.0.downsample.1.running_var), scope: LResNet2/Sequential[layer3]/BlockIR2[0]/Sequential[downsample]/BatchNorm2d[1]
# /home/huangry/anaconda3/lib/python3.7/site-packages/torch/nn/functional.py:1670:0
     %167 : Float(1, 128, 14, 14) = onnx::Add(%164, %166), scope: LResNet2/Sequential[layer3]/BlockIR2[0]
# /home/huangry/program/LResNet18E.py:44:0
     %168 : Float(1, 128, 14, 14) = onnx::Conv[dilations=[1, 1], group=1, kernel_shape=[3,
3], pads=[1, 1, 1, 1], strides=[1, 1]](%167, %layer3.1.conv1.weight), scope: LResNet2/Sequential[layer3]/BlockIR2[1]/Conv2d[conv1]
# /home/huangry/anaconda3/lib/python3.7/site-packages/torch/nn/modules/conv.py:342:0
     %169 : Float(1, 128, 14, 14) = onnx::BatchNormalization[epsilon=1e-05, momentum=0.9](%168,
%layer3.1.bn2.weight, %layer3.1.bn2.bias, %layer3.1.bn2.running_mean, %layer3.1.bn2.running_var),
scope: LResNet2/Sequential[layer3]/BlockIR2[1]/BatchNorm2d[bn2] # /home/huangry/anaconda3/lib/python3.7/site-packages/torch/nn/functional.py:1670:0
     %170 : Float(1, 128, 14, 14) = onnx::Relu(%169), scope: LResNet2/Sequential[layer3]/BlockIR2[1]/ReLU[ReLU1]
# /home/huangry/anaconda3/lib/python3.7/site-packages/torch/nn/functional.py:912:0
     %171 : Float(1, 128, 14, 14) = onnx::Conv[dilations=[1, 1], group=1, kernel_shape=[3,
3], pads=[1, 1, 1, 1], strides=[1, 1]](%170, %layer3.1.conv2.weight), scope: LResNet2/Sequential[layer3]/BlockIR2[1]/Conv2d[conv2]
# /home/huangry/anaconda3/lib/python3.7/site-packages/torch/nn/modules/conv.py:342:0
     %172 : Float(1, 128, 14, 14) = onnx::BatchNormalization[epsilon=1e-05, momentum=0.9](%171,
%layer3.1.bn3.weight, %layer3.1.bn3.bias, %layer3.1.bn3.running_mean, %layer3.1.bn3.running_var),
scope: LResNet2/Sequential[layer3]/BlockIR2[1]/BatchNorm2d[bn3] # /home/huangry/anaconda3/lib/python3.7/site-packages/torch/nn/functional.py:1670:0
     %173 : Float(1, 128, 14, 14) = onnx::Add(%172, %167), scope: LResNet2/Sequential[layer3]/BlockIR2[1]
# /home/huangry/program/LResNet18E.py:44:0
     %174 : Float(1, 384, 14, 14) = onnx::Conv[dilations=[1, 1], group=1, kernel_shape=[3,
3], pads=[1, 1, 1, 1], strides=[1, 1]](%173, %layer4.0.conv1.weight), scope: LResNet2/Sequential[layer4]/BlockIR2[0]/Conv2d[conv1]
# /home/huangry/anaconda3/lib/python3.7/site-packages/torch/nn/modules/conv.py:342:0
     %175 : Float(1, 384, 14, 14) = onnx::BatchNormalization[epsilon=1e-05, momentum=0.9](%174,
%layer4.0.bn2.weight, %layer4.0.bn2.bias, %layer4.0.bn2.running_mean, %layer4.0.bn2.running_var),
scope: LResNet2/Sequential[layer4]/BlockIR2[0]/BatchNorm2d[bn2] # /home/huangry/anaconda3/lib/python3.7/site-packages/torch/nn/functional.py:1670:0
     %176 : Float(1, 384, 14, 14) = onnx::Relu(%175), scope: LResNet2/Sequential[layer4]/BlockIR2[0]/ReLU[ReLU1]
# /home/huangry/anaconda3/lib/python3.7/site-packages/torch/nn/functional.py:912:0
     %177 : Float(1, 384, 7, 7) = onnx::Conv[dilations=[1, 1], group=1, kernel_shape=[3, 3],
pads=[1, 1, 1, 1], strides=[2, 2]](%176, %layer4.0.conv2.weight), scope: LResNet2/Sequential[layer4]/BlockIR2[0]/Conv2d[conv2]
# /home/huangry/anaconda3/lib/python3.7/site-packages/torch/nn/modules/conv.py:342:0
     %178 : Float(1, 384, 7, 7) = onnx::BatchNormalization[epsilon=1e-05, momentum=0.9](%177,
%layer4.0.bn3.weight, %layer4.0.bn3.bias, %layer4.0.bn3.running_mean, %layer4.0.bn3.running_var),
scope: LResNet2/Sequential[layer4]/BlockIR2[0]/BatchNorm2d[bn3] # /home/huangry/anaconda3/lib/python3.7/site-packages/torch/nn/functional.py:1670:0
     %179 : Float(1, 384, 7, 7) = onnx::Conv[dilations=[1, 1], group=1, kernel_shape=[1, 1],
pads=[0, 0, 0, 0], strides=[2, 2]](%173, %layer4.0.downsample.0.weight), scope: LResNet2/Sequential[layer4]/BlockIR2[0]/Sequential[downsample]/Conv2d[0]
# /home/huangry/anaconda3/lib/python3.7/site-packages/torch/nn/modules/conv.py:342:0
     %180 : Float(1, 384, 7, 7) = onnx::BatchNormalization[epsilon=1e-05, momentum=0.9](%179,
%layer4.0.downsample.1.weight, %layer4.0.downsample.1.bias, %layer4.0.downsample.1.running_mean,
%layer4.0.downsample.1.running_var), scope: LResNet2/Sequential[layer4]/BlockIR2[0]/Sequential[downsample]/BatchNorm2d[1]
# /home/huangry/anaconda3/lib/python3.7/site-packages/torch/nn/functional.py:1670:0
     %181 : Float(1, 384, 7, 7) = onnx::Add(%178, %180), scope: LResNet2/Sequential[layer4]/BlockIR2[0]
# /home/huangry/program/LResNet18E.py:44:0
     %182 : Float(1, 384, 7, 7) = onnx::Conv[dilations=[1, 1], group=1, kernel_shape=[3, 3],
pads=[1, 1, 1, 1], strides=[1, 1]](%181, %layer4.1.conv1.weight), scope: LResNet2/Sequential[layer4]/BlockIR2[1]/Conv2d[conv1]
# /home/huangry/anaconda3/lib/python3.7/site-packages/torch/nn/modules/conv.py:342:0
     %183 : Float(1, 384, 7, 7) = onnx::BatchNormalization[epsilon=1e-05, momentum=0.9](%182,
%layer4.1.bn2.weight, %layer4.1.bn2.bias, %layer4.1.bn2.running_mean, %layer4.1.bn2.running_var),
scope: LResNet2/Sequential[layer4]/BlockIR2[1]/BatchNorm2d[bn2] # /home/huangry/anaconda3/lib/python3.7/site-packages/torch/nn/functional.py:1670:0
     %184 : Float(1, 384, 7, 7) = onnx::Relu(%183), scope: LResNet2/Sequential[layer4]/BlockIR2[1]/ReLU[ReLU1]
# /home/huangry/anaconda3/lib/python3.7/site-packages/torch/nn/functional.py:912:0
     %185 : Float(1, 384, 7, 7) = onnx::Conv[dilations=[1, 1], group=1, kernel_shape=[3, 3],
pads=[1, 1, 1, 1], strides=[1, 1]](%184, %layer4.1.conv2.weight), scope: LResNet2/Sequential[layer4]/BlockIR2[1]/Conv2d[conv2]
# /home/huangry/anaconda3/lib/python3.7/site-packages/torch/nn/modules/conv.py:342:0
     %186 : Float(1, 384, 7, 7) = onnx::BatchNormalization[epsilon=1e-05, momentum=0.9](%185,
%layer4.1.bn3.weight, %layer4.1.bn3.bias, %layer4.1.bn3.running_mean, %layer4.1.bn3.running_var),
scope: LResNet2/Sequential[layer4]/BlockIR2[1]/BatchNorm2d[bn3] # /home/huangry/anaconda3/lib/python3.7/site-packages/torch/nn/functional.py:1670:0
     %187 : Float(1, 384, 7, 7) = onnx::Add(%186, %181), scope: LResNet2/Dropout[drop] # /home/huangry/anaconda3/lib/python3.7/site-packages/torch/nn/functional.py:807:0
     %188 : Tensor = onnx::Pad[mode="constant", pads=[0, 0, 0, 0, 0, 0, 0, 0], value=0](%187),
scope: LResNet2/AvgPool2d[avgpool]
     %189 : Float(1, 384, 1, 1) = onnx::AveragePool[kernel_shape=[7, 7], pads=[0, 0, 0, 0],
strides=[1, 1]](%188), scope: LResNet2/AvgPool2d[avgpool] # /home/huangry/anaconda3/lib/python3.7/site-packages/torch/nn/modules/pooling.py:554:0
     %190 : Long() = onnx::Constant[value={0}](), scope: LResNet2
     %191 : Tensor = onnx::Shape(%189), scope: LResNet2
     %192 : Long() = onnx::Gather[axis=0](%191, %190), scope: LResNet2 # /home/huangry/program/LResNet18E.py:104:0
     %193 : Long() = onnx::Constant[value={-1}](), scope: LResNet2
     %194 : Tensor = onnx::Unsqueeze[axes=[0]](%192)
     %195 : Tensor = onnx::Unsqueeze[axes=[0]](%193)
     %196 : Tensor = onnx::Concat[axis=0](%194, %195)
     %197 : Float(1, 384) = onnx::Reshape(%189, %196), scope: LResNet2 # /home/huangry/program/LResNet18E.py:104:0
     %198 : Float(1, 8) = onnx::Gemm[alpha=1, beta=1, transB=1](%197, %classifier.weight,
%classifier.bias), scope: LResNet2/Linear[classifier] # /home/anaconda3/lib/python3.7/site-packages/torch/nn/functional.py:1370:0
     return (%198)`
   

----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
 
For queries about this service, please contact Infrastructure at:
users@infra.apache.org


With regards,
Apache Git Services

Mime
View raw message