Pytorch--OverflowError: cannot serialize a bytes object larger than 4 GiB

最新推荐文章于 2024-07-17 16:02:09 发布

_夜风

最新推荐文章于 2024-07-17 16:02:09 发布

阅读量7.3k

点赞数 3

分类专栏： python

本文链接：https://blog.csdn.net/k183000860/article/details/107624570

版权

本文介绍了在使用Pytorch时遇到的两种错误：OverflowError（无法序列化超过4GiB的字节对象）和CUDA out of memory。针对前者，解决方案是将num_workers设置为0，这能解决Windows上的问题；对于后者，可以通过减小batch_size或者更换GPU来缓解内存不足的问题。

摘要由CSDN通过智能技术生成

首先给出报错提示

[07.27.20|23:44:32] Training epoch: 0
Traceback (most recent call last):
  File "main.py", line 31, in <module>
    p.start()
  File "D:\code\st-gcn\processor\processor.py", line 114, in start
    self.train()
  File "D:\code\st-gcn\processor\recognition.py", line 85, in train
    for data, label in loader:
  File "D:\Python36\lib\site-packages\torch\utils\data\dataloader.py", line 279, in __iter__
    return _MultiProcessingDataLoaderIter(self)
  File "D:\Python36\lib\site-packages\torch\utils\data\dataloader.py", line 719, in __init__
    w.start()
  File "D:\Python36\lib\multiprocessing\process.py", line 105, in start
    self._popen = self._Popen(self)
  File "D:\Python36\lib\multiprocessing\context.py", line 223, in _Popen
    return _default_context.get_context().Process._Popen(process_obj)
  File "D:\Python36\lib\multiprocessing\context.py", line 322, in