1. 在使用DataLoader加载数据时报错：OSError: [Errno 22] Invalid argument

for i, data in enumerate(loader):

并且这个错误在数据量小的时候没有发生，在大数据量下才产生。我的办法是将DataLoader的参数num_workers由1改为0，程序正常运行。

2. 在给网络输入Tensor时报错：RuntimeError: Expected object of scalar type Double but got scalar type Float for argument #2 ‘mat2’ in call to _th_mm

按照提示是数据类型的原因，查看输入数据类型为torch.float64，将输入数据类型转为torch.float32，搞定。

3. 使用view时报错：RuntimeError: view size is not compatible with input tensor’s size and stride (at least one dimension spans across two contiguous subspaces). Use .reshape(…) instead.

tensor不是contiguous 引起的错误。查看_targets.is_contiguous()为False。
两种解决办法：1）按照提示使用reshape代替；2）将变量先转为_contiguous ，再进行view:

targets.contiguous().view(targets.size(0)*targets.size(1),-1)

4. 使用交叉熵损失函数报错：Expected object of type torch.LongTensor but found type torch.FloatTensor for argument #2 ‘target’

数据类型不匹配的原因，按照提示，将目标值类型转为torch.Long。

5. 还是使用交叉熵损失函数报错：multi-target not supported at C:\w\1\s\windows\pytorch\aten\src\THNN/generic/ClassNLLCriterion.c:21

原因是torch.nn.CrossEntropyLoss()接受的目标值必须是类标值，而不是one-hot编码，将目标值改为类标值即可。

https://zhuanlan.zhihu.com/p/112099126

6.Segmentation fault (core dumped)

7.OSError: image file is truncated (16 bytes not processed)

Traceback (most recent call last):
2012   File "train.py", line 131, in <module>
2013     for _, (input_images, ground_truths, masks) in enumerate(data_loader):
2014   File "/data/guoxiefan/anaconda3/envs/pytorch/lib/python3.6/site-packages/torch/utils/data/dataloader.py", line 345, in __next__
2015     data = self._next_data()
2016   File "/data/guoxiefan/anaconda3/envs/pytorch/lib/python3.6/site-packages/torch/utils/data/dataloader.py", line 856, in _next_data
2017     return self._process_data(data)
2018   File "/data/guoxiefan/anaconda3/envs/pytorch/lib/python3.6/site-packages/torch/utils/data/dataloader.py", line 881, in _process_data
2019     data.reraise()
2020   File "/data/guoxiefan/anaconda3/envs/pytorch/lib/python3.6/site-packages/torch/_utils.py", line 394, in reraise
2021     raise self.exc_type(msg)
2022 OSError: Caught OSError in DataLoader worker process 3.
2023 Original Traceback (most recent call last):
2024   File "/data/guoxiefan/anaconda3/envs/pytorch/lib/python3.6/site-packages/torch/utils/data/_utils/worker.py", line 178, in _worker_loop
2025     data = fetcher.fetch(index)
2026   File "/data/guoxiefan/anaconda3/envs/pytorch/lib/python3.6/site-packages/torch/utils/data/_utils/fetch.py", line 44, in fetch
2027     data = [self.dataset[idx] for idx in possibly_batched_index]
2028   File "/data/guoxiefan/anaconda3/envs/pytorch/lib/python3.6/site-packages/torch/utils/data/_utils/fetch.py", line 44, in <listcomp>
2029     data = [self.dataset[idx] for idx in possibly_batched_index]
2030   File "/home/guoxiefan/PyTorch/ImageInpainting/LBAM/src/dataset.py", line 76, in __getitem__
2031     ground_truth = self.image_files_transforms(image.convert('RGB'))
2032   File "/data/guoxiefan/anaconda3/envs/pytorch/lib/python3.6/site-packages/PIL/Image.py", line 873, in convert
2033     self.load()
2034   File "/data/guoxiefan/anaconda3/envs/pytorch/lib/python3.6/site-packages/PIL/ImageFile.py", line 247, in load
2035     "(%d bytes not processed)" % len(b)
2036 OSError: image file is truncated (16 bytes not processed)

解决方案：[Link]

It’s probably you installed Pillow instead of PIL. Try adding these todatasets.pyfile at line 14:
from PIL import ImageFileImageFile.LOAD_TRUNCATED_IMAGES = True

8RuntimeError: cuda runtime error (60) : peer mapping resources exhausted at /opt/conda/conda-bld/pytorch_1579022051443/work/aten/src/THC/THCGeneral.cpp:141

Traceback (most recent call last):
  File "train.py", line 136, in <module>
    outputs = generator(input_images, masks)
  File "/data/guoxiefan/anaconda3/envs/pytorch/lib/python3.6/site-packages/torch/nn/modules/module.py", line 532, in __call__
    result = self.forward(*input, **kwargs)
  File "/data/guoxiefan/anaconda3/envs/pytorch/lib/python3.6/site-packages/torch/nn/parallel/data_parallel.py", line 148, in forward
    inputs, kwargs = self.scatter(inputs, kwargs, self.device_ids)
  File "/data/guoxiefan/anaconda3/envs/pytorch/lib/python3.6/site-packages/torch/nn/parallel/data_parallel.py", line 159, in scatter
    return scatter_kwargs(inputs, kwargs, device_ids, dim=self.dim)
  File "/data/guoxiefan/anaconda3/envs/pytorch/lib/python3.6/site-packages/torch/nn/parallel/scatter_gather.py", line 36, in scatter_kwargs
    inputs = scatter(inputs, target_gpus, dim) if inputs else []
  File "/data/guoxiefan/anaconda3/envs/pytorch/lib/python3.6/site-packages/torch/nn/parallel/scatter_gather.py", line 28, in scatter
    res = scatter_map(inputs)
  File "/data/guoxiefan/anaconda3/envs/pytorch/lib/python3.6/site-packages/torch/nn/parallel/scatter_gather.py", line 15, in scatter_map
    return list(zip(*map(scatter_map, obj)))
  File "/data/guoxiefan/anaconda3/envs/pytorch/lib/python3.6/site-packages/torch/nn/parallel/scatter_gather.py", line 13, in scatter_map
    return Scatter.apply(target_gpus, None, dim, obj)
  File "/data/guoxiefan/anaconda3/envs/pytorch/lib/python3.6/site-packages/torch/nn/parallel/_functions.py", line 89, in forward
    outputs = comm.scatter(input, target_gpus, chunk_sizes, ctx.dim, streams)
  File "/data/guoxiefan/anaconda3/envs/pytorch/lib/python3.6/site-packages/torch/cuda/comm.py", line 147, in scatter
    return tuple(torch._C._scatter(tensor, devices, chunk_sizes, dim, streams))
RuntimeError: cuda runtime error (60) : peer mapping resources exhausted at /opt/conda/conda-bld/pytorch_1579022051443/work/aten/src/THC/THCGeneral.cpp:141

解决：对于nn.DataParallel作用的nn.Module，传入参数一般为实数，或者为原始数据（[B C H * W]）。传入参数与并行有关，需要特别注意。nn.DataParallel并行切分 B 维度。

9.RuntimeError: received 0 items of ancdata

Traceback (most recent call last):
  File "train.py", line 123, in <module>
    for _, (input_images, ground_truths, masks) in enumerate(data_loader):
  File "/data/guoxiefan/anaconda3/envs/pytorch/lib/python3.6/site-packages/torch/utils/data/dataloader.py", line 345, in __next__
    data = self._next_data()
  File "/data/guoxiefan/anaconda3/envs/pytorch/lib/python3.6/site-packages/torch/utils/data/dataloader.py", line 841, in _next_data
    idx, data = self._get_data()
  File "/data/guoxiefan/anaconda3/envs/pytorch/lib/python3.6/site-packages/torch/utils/data/dataloader.py", line 808, in _get_data
    success, data = self._try_get_data()
  File "/data/guoxiefan/anaconda3/envs/pytorch/lib/python3.6/site-packages/torch/utils/data/dataloader.py", line 761, in _try_get_data
    data = self._data_queue.get(timeout=timeout)
  File "/data/guoxiefan/anaconda3/envs/pytorch/lib/python3.6/multiprocessing/queues.py", line 113, in get
    return _ForkingPickler.loads(res)
  File "/data/guoxiefan/anaconda3/envs/pytorch/lib/python3.6/site-packages/torch/multiprocessing/reductions.py", line 294, in rebuild_storage_fd
    fd = df.detach()
  File "/data/guoxiefan/anaconda3/envs/pytorch/lib/python3.6/multiprocessing/resource_sharer.py", line 58, in detach
    return reduction.recv_handle(conn)
  File "/data/guoxiefan/anaconda3/envs/pytorch/lib/python3.6/multiprocessing/reduction.py", line 182, in recv_handle
    return recvfds(s, 1)[0]
  File "/data/guoxiefan/anaconda3/envs/pytorch/lib/python3.6/multiprocessing/reduction.py", line 161, in recvfds
    len(ancdata))
RuntimeError: received 0 items of ancdata

设置加载数据的num_workers=0,
可查看，当需共享的tensor超过open files限制时，即会出现该错误。
解决办法有2种：

1、增加open files的限制数量：
不能用sudo ulimit -n命令，而需执行：

sudo sh -c “ulimit -n 65535 && exec su $LOGNAME”
2、修改多线程的tensor方式为file_system（默认方式为file_descriptor，受限于open files数量）：
torch.multiprocessing.set_sharing_strategy(‘file_system’)

解决 [Link] [Link]

Traceback (most recent call last):
  File "train.py", line 124, in <module>
  File "/data/guoxiefan/anaconda3/envs/pytorch/lib/python3.6/site-packages/torch/utils/data/dataloader.py", line 345, in __next__
  File "/data/guoxiefan/anaconda3/envs/pytorch/lib/python3.6/site-packages/torch/utils/data/dataloader.py", line 841, in _next_data
  File "/data/guoxiefan/anaconda3/envs/pytorch/lib/python3.6/site-packages/torch/utils/data/dataloader.py", line 808, in _get_data
  File "/data/guoxiefan/anaconda3/envs/pytorch/lib/python3.6/site-packages/torch/utils/data/dataloader.py", line 761, in _try_get_data
  File "/data/guoxiefan/anaconda3/envs/pytorch/lib/python3.6/multiprocessing/queues.py", line 113, in get
  File "/data/guoxiefan/anaconda3/envs/pytorch/lib/python3.6/site-packages/torch/multiprocessing/reductions.py", line 310, in rebuild_storage_filename
RuntimeError: unable to open shared memory object </torch_24388_2219814394> in read-write mode

Pytorch API 使用文档 - 官方教程

pytorch error