基于Pytorch(version=1.0)框架,利用torch.nn.LSTM模块构建对于时间序列数据的分类器。
训练过程中,当网络进行反向传播时,程序曝出异常
Traceback (most recent call last):
File "train_version1.py", line 236, in <module>
loss.backward()
File "D:\ProgramFiles\Anaconda\lib\site-packages\torch\tensor.py", line 102, in backward
torch.autograd.backward(self, gradient, retain_graph, create_graph)
File "D:\ProgramFiles\Anaconda\lib\site-packages\torch\autograd\__init__.py", line 90, in backward
allow_unreachable=True) # allow_unreachable flag
RuntimeError: cur_offset == offset ASSERT FAILED at ..\aten\src\ATen\native\cudnn\RNN.cpp:471, please report a bug to PyTorch. cur_offset = 1860; offset = 930
刚看到这个异常我是懵的,因为,显然错误出现在反向传播这一块,看起来并不是代码逻辑出了问题
经过一番查阅,在GitHub上有一位东京小哥的帖子引起了我的注意--