Python GANet的安装

我正在尝试使用以下NN：

当我不使用apex时会发生此错误

Traceback (most recent call last):
  File "predict.py", line 187, in <module>
    test(leftname, rightname, savename)
  File "predict.py", line 158, in test
    prediction = model(input1, input2)
  File "/home/master/anaconda3/lib/python3.7/site-packages/torch/nn/modules/module.py", line 489, in __call__
    result = self.forward(*input, **kwargs)
  File "/home/master/anaconda3/lib/python3.7/site-packages/torch/nn/parallel/data_parallel.py", line 141, in forward
    return self.module(*inputs[0], **kwargs[0])
  File "/home/master/anaconda3/lib/python3.7/site-packages/torch/nn/modules/module.py", line 489, in __call__
    result = self.forward(*input, **kwargs)
  File "/home/master/Work/Project01/GANet/models/GANet_deep.py", line 396, in forward
    g = self.conv_start(x)  
  File "/home/master/anaconda3/lib/python3.7/site-packages/torch/nn/modules/module.py", line 489, in __call__
    result = self.forward(*input, **kwargs)
  File "/home/master/anaconda3/lib/python3.7/site-packages/torch/nn/modules/container.py", line 92, in forward
    input = module(input)
  File "/home/master/anaconda3/lib/python3.7/site-packages/torch/nn/modules/module.py", line 489, in __call__
    result = self.forward(*input, **kwargs)
  File "/home/master/Work/Project01/GANet/models/GANet_deep.py", line 42, in forward
    x = self.bn(x)
  File "/home/master/anaconda3/lib/python3.7/site-packages/torch/nn/modules/module.py", line 489, in __call__
    result = self.forward(*input, **kwargs)
  File "/home/master/Work/Project01/GANet/libs/sync_bn/modules/sync_bn.py", line 121, in forward
    self.activation, self.slope).view(input_shape)
  File "/home/master/Work/Project01/GANet/libs/sync_bn/functions/sync_bn.py", line 95, in forward
    y = sync_bn_gpu.batchnorm_forward(x, _ex, _exs, gamma, beta, ctx.eps)
RuntimeError: cudaGetLastError() == cudaSuccess ASSERT FAILED at src/gpu/sync_bn_cuda.cu:289, please report a bug to PyTorch. (BatchNorm_Forward_CUDA at src/gpu/sync_bn_cuda.cu:289)
frame #0: c10::Error::Error(c10::SourceLocation, std::string const&) + 0x45 (0x7fca7ac1bcc5 in /home/master/anaconda3/lib/python3.7/site-packages/torch/lib/libc10.so)
frame #1: BatchNorm_Forward_CUDA(at::Tensor, at::Tensor, at::Tensor, at::Tensor, at::Tensor, float) + 0x2be (0x7fca76e56ff1 in /home/master/Work/Project01/GANet/libs/sync_bn/build/lib/sync_bn_gpu.cpython-37m-x86_64-linux-gnu.so)
frame #2: <unknown function> + 0x22ab3 (0x7fca76e54ab3 in /home/master/Work/Project01/GANet/libs/sync_bn/build/lib/sync_bn_gpu.cpython-37m-x86_64-linux-gnu.so)
frame #3: <unknown function> + 0x201a5 (0x7fca76e521a5 in /home/master/Work/Project01/GANet/libs/sync_bn/build/lib/sync_bn_gpu.cpython-37m-x86_64-linux-gnu.so)
<omitting python frames>
frame #11: THPFunction_apply(_object*, _object*) + 0x5a1 (0x7fcaab0b4bf1 in /home/master/anaconda3/lib/python3.7/site-packages/torch/lib/libtorch_python.so)

apex（并将所有BatchNormal2D和BatchNormal3D替换为apex.parallel.syncBatchNormal）运行NN，但返回所有零结果矩阵。 有什么建议吗

在ubuntu 18.04上工作 和康达在一起， python 3.7.4， 火炬1.0.0， cuda 10.2

gcc 7.5

pip uninstall apex
rm -rf apex

git clone https://github.com/ptrblck/apex.git
cd apex
git checkout apex_no_distributed
pip install -v --no-cache-dir ./