深度学习GitHub复现常见错误之参数num_workers引发的错误：“BrokenPipeError: [Errno 32] Broken pipe”

简单记录下这个错误，不止遇到一次了，还好我记得怎么回事，错误如下：

C:\ProgramData\Anaconda3\lib\site-packages\h5py\__init__.py:36: FutureWarning: Conversion of the second argument of issubdtype from `float` to `np.floating` is deprecated. In future, it will be treated as `np.float64 == np.dtype(float).type`.
  from ._conv import register_converters as _register_converters
Files already downloaded and verified
INFO: Starting training from global step 0...
C:\ProgramData\Anaconda3\lib\site-packages\h5py\__init__.py:36: FutureWarning: Conversion of the second argument of issubdtype from `float` to `np.floating` is deprecated. In future, it will be treated as `np.float64 == np.dtype(float).type`.
  from ._conv import register_converters as _register_converters
Files already downloaded and verified
INFO: Starting training from global step 0...
Traceback (most recent call last):
  File "", line 1, in
  File "C:\ProgramData\Anaconda3\lib\multiprocessing\spawn.py", line 105, in spawn_main
    exitcode = _main(fd)
  File "C:\ProgramData\Anaconda3\lib\multiprocessing\spawn.py", line 114, in _main
    prepare(preparation_data)
  File "C:\ProgramData\Anaconda3\lib\multiprocessing\spawn.py", line 225, in prepare
    _fixup_main_from_path(data['init_main_from_path'])
  File "C:\ProgramData\Anaconda3\lib\multiprocessing\spawn.py", line 277, in _fixup_main_from_path
    run_name="__mp_main__")
  File "C:\ProgramData\Anaconda3\lib\runpy.py", line 263, in run_path
    pkg_name=pkg_name, script_name=fname)
  File "C:\ProgramData\Anaconda3\lib\runpy.py", line 96, in _run_module_code
    mod_name, mod_spec, pkg_name, script_name)
  File "C:\ProgramData\Anaconda3\lib\runpy.py", line 85, in _run_code
    exec(code, run_globals)
  File "D:\wcs\GAN\train.py", line 30, in
    trainer.train()
  File "C:\Users\tq\AppData\Roaming\Python\Python36\site-packages\torch_mimicry\training\trainer.py", line 239, in train
    iter_dataloader = iter(self.dataloader)
  File "C:\ProgramData\Anaconda3\lib\site-packages\torch\utils\data\dataloader.py", line 193, in __iter__
    return _DataLoaderIter(self)
  File "C:\ProgramData\Anaconda3\lib\site-packages\torch\utils\data\dataloader.py", line 469, in __init__
    w.start()
  File "C:\ProgramData\Anaconda3\lib\multiprocessing\process.py", line 105, in start
    self._popen = self._Popen(self)
  File "C:\ProgramData\Anaconda3\lib\multiprocessing\context.py", line 223, in _Popen
    return _default_context.get_context().Process._Popen(process_obj)
  File "C:\ProgramData\Anaconda3\lib\multiprocessing\context.py", line 322, in _Popen
    return Popen(process_obj)
  File "C:\ProgramData\Anaconda3\lib\multiprocessing\popen_spawn_win32.py", line 33, in __init__
    prep_data = spawn.get_preparation_data(process_obj._name)
  File "C:\ProgramData\Anaconda3\lib\multiprocessing\spawn.py", line 143, in get_preparation_data
    _check_not_importing_main()
  File "C:\ProgramData\Anaconda3\lib\multiprocessing\spawn.py", line 136, in _check_not_importing_main
    is not going to be frozen to produce an executable.''')
RuntimeError:
        An attempt has been made to start a new process before the
        current process has finished its bootstrapping phase.

        This probably means that you are not using fork to start your
        child processes and you have forgotten to use the proper idiom
        in the main module:

            if __name__ == '__main__':
                freeze_support()
                ...

        The "freeze_support()" line can be omitted if the program
        is not going to be frozen to produce an executable.
Traceback (most recent call last):
  File "train.py", line 30, in
    trainer.train()
  File "C:\Users\tq\AppData\Roaming\Python\Python36\site-packages\torch_mimicry\training\trainer.py", line 239, in train
    iter_dataloader = iter(self.dataloader)
  File "C:\ProgramData\Anaconda3\lib\site-packages\torch\utils\data\dataloader.py", line 193, in __iter__
    return _DataLoaderIter(self)
  File "C:\ProgramData\Anaconda3\lib\site-packages\torch\utils\data\dataloader.py", line 469, in __init__
    w.start()
  File "C:\ProgramData\Anaconda3\lib\multiprocessing\process.py", line 105, in start
    self._popen = self._Popen(self)
  File "C:\ProgramData\Anaconda3\lib\multiprocessing\context.py", line 223, in _Popen
    return _default_context.get_context().Process._Popen(process_obj)
  File "C:\ProgramData\Anaconda3\lib\multiprocessing\context.py", line 322, in _Popen
    return Popen(process_obj)
  File "C:\ProgramData\Anaconda3\lib\multiprocessing\popen_spawn_win32.py", line 65, in __init__
    reduction.dump(process_obj, to_child)
  File "C:\ProgramData\Anaconda3\lib\multiprocessing\reduction.py", line 60, in dump
    ForkingPickler(file, protocol).dump(obj)
BrokenPipeError: [Errno 32] Broken pipe

 

这个是我测试GAN网络时遇到的，框架是pytorch，但是说明一下，不只是这一种网络和这一种框架会遇到这个问题，是所有的框架和网络都会遇到的问题，因为这算是一个跨平台的错误，也就是项目在Linux可以运行，但是到了Windows就会报错，引起错误的参数是num_workers，这是一个控制进程数的参数，代码片段如下：

device = torch.device('cuda:0' if torch.cuda.is_available() else "cpu")
dataset = mmc.datasets.load_dataset(root='./datasets', name='cifar10')
dataloader = torch.utils.data.DataLoader(
    dataset, batch_size=64, shuffle=True, num_workers=4)

请记住，遇到这个问题第一反应都应该是把num_workers改为等于0，不行了再进一步看。