关于isic2017shujuji

[lijincan11]

您好作者，您的研究非常的棒，但是我在复现的过程中出现了一点问题，训练完isic2017数据集后，进行测试的时候出现了如下的问题 ../aten/src/ATen/native/cuda/Loss.cu:92: operator(): block: [71,0,0], thread: [27,0,0] Assertion input_val >= zero && input_val <= one failed. ../aten/src/ATen/native/cuda/Loss.cu:92: operator(): block: [71,0,0], thread: [28,0,0] Assertion input_val >= zero && input_val <= one failed. ../aten/src/ATen/native/cuda/Loss.cu:92: operator(): block: [71,0,0], thread: [29,0,0] Assertion input_val >= zero && input_val <= one failed. ../aten/src/ATen/native/cuda/Loss.cu:92: operator(): block: [71,0,0], thread: [30,0,0] Assertion input_val >= zero && input_val <= one failed. ../aten/src/ATen/native/cuda/Loss.cu:92: operator(): block: [71,0,0], thread: [31,0,0] Assertion input_val >= zero && input_val <= one failed. Clipping input data to the valid range for imshow with RGB data ([0..1] for floats or [0..255] for integers). 17%|██████████████████████████████████▋ | 101/600 [00:08<00:42, 11.78it/s] Traceback (most recent call last): File "UltraLight-VM-UNet_17/train.py", line 171, in main(config) File "UltraLight-VM-UNet_17/train.py", line 156, in main loss = test_one_epoch( File "/home/ljc/UltraLight-VM-UNet_17/engine.py", line 122, in test_one_epoch out = model(img) File "/home/xjl/anaconda3/envs/ulvm/lib/python3.8/site-packages/torch/nn/modules/module.py", line 1190, in _call_impl return forward_call(*input, kwargs) File "/home/xjl/anaconda3/envs/ulvm/lib/python3.8/site-packages/torch/nn/parallel/data_parallel.py", line 169, in forward return self.module(*inputs[0], *kwargs[0]) File "/home/xjl/anaconda3/envs/ulvm/lib/python3.8/site-packages/torch/nn/modules/module.py", line 1190, in _call_impl return forward_call(input, kwargs) File "/home/ljc/UltraLight-VM-UNet_17/models/UltraLight_VM_UNet.py", line 210, in forward out = F.gelu(F.max_pool2d(self.ebn1(self.encoder1(x)),2,2)) File "/home/xjl/anaconda3/envs/ulvm/lib/python3.8/site-packages/torch/nn/modules/module.py", line 1190, in _call_impl return forward_call(*input, *kwargs) File "/home/xjl/anaconda3/envs/ulvm/lib/python3.8/site-packages/torch/nn/modules/container.py", line 204, in forward input = module(input) File "/home/xjl/anaconda3/envs/ulvm/lib/python3.8/site-packages/torch/nn/modules/module.py", line 1190, in _call_impl return forward_call(input, **kwargs) File "/home/xjl/anaconda3/envs/ulvm/lib/python3.8/site-packages/torch/nn/modules/conv.py", line 463, in forward return self._conv_forward(input, self.weight, self.bias) File "/home/xjl/anaconda3/envs/ulvm/lib/python3.8/site-packages/torch/nn/modules/conv.py", line 459, in _conv_forward return F.conv2d(input, weight, bias, self.stride, RuntimeError: cuDNN error: CUDNN_STATUS_MAPPING_ERROR

[wurenkai]

Hi! Try restarting your environment and check if the output goes through a sigmoid activation function (return torch.sigmoid(out0)).

[heitorrapela]

Hello, thanks for the work. I have the same error; I printed the output, and it is giving nan:

[heitorrapela]

I think that maybe the problem is in the Prepare_ISIC2017.py, we have to change it for the actual dataset provided on the README.md

[heitorrapela]

I fixed that installing the correct versions of the libraries for the preprocessing, then the data was okay to train, if someone have the same issue, here is the other thread about it: Issue 4

[wurenkai]

Yes, thank you. Whether the data is loaded correctly is also a key issue.