关于stack expects a non-empty TensorList的问题，又要叨扰下作者。 about pytorch_bert_bilstm_crf_ner HOT 6 CLOSED

我今天也出现了这种错误，最后我把所有的输出形状都打印了出来，发现最后一个batch_size的大小为1，所以导致了这种错误，只要把dataloader的drop_last设置为True，就能正常运行了

from pytorch_bert_bilstm_crf_ner.

> E:\env\py7\lib\site-packages\transformers\optimization.py:309: FutureWarning: This implementation of AdamW is deprecated and will be removed in a future version. Use thePyTorch implementation torch.optim.AdamW instead, or set `no_deprecation_warning=True` to disable this warning
>   `FutureWarning,`
> Traceback (most recent call last):
>   File "F:/pytorch_bert_bilstm_crf_ner-main/pytorch_bert_bilstm_crf_ner-main/main.py", line 214, in <module>
>     bertForNer.train()
>   File "F:/pytorch_bert_bilstm_crf_ner-main/pytorch_bert_bilstm_crf_ner-main/main.py", line 46, in train
>     torch.nn.utils.clip_grad_norm_(self.model.parameters(), self.args.max_grad_norm)
>   File "E:\env\py7\lib\site-packages\torch\nn\utils\clip_grad.py", line 31, in clip_grad_norm_
>     total_norm = torch.norm(torch.stack([torch.norm(p.grad.detach(), norm_type) for p in parameters]), norm_type)
> RuntimeError: stack expects a non-empty TensorList

> 

这是我运行出来的错误

from pytorch_bert_bilstm_crf_ner.

> E:\env\py7\lib\site-packages\transformers\optimization.py:309: FutureWarning: This implementation of AdamW is deprecated and will be removed in a future version. Use thePyTorch implementation torch.optim.AdamW instead, or set `no_deprecation_warning=True` to disable this warning
>   `FutureWarning,`
> Traceback (most recent call last):
>   File "F:/pytorch_bert_bilstm_crf_ner-main/pytorch_bert_bilstm_crf_ner-main/main.py", line 214, in <module>
>     bertForNer.train()
>   File "F:/pytorch_bert_bilstm_crf_ner-main/pytorch_bert_bilstm_crf_ner-main/main.py", line 46, in train
>     torch.nn.utils.clip_grad_norm_(self.model.parameters(), self.args.max_grad_norm)
>   File "E:\env\py7\lib\site-packages\torch\nn\utils\clip_grad.py", line 31, in clip_grad_norm_
>     total_norm = torch.norm(torch.stack([torch.norm(p.grad.detach(), norm_type) for p in parameters]), norm_type)
> RuntimeError: stack expects a non-empty TensorList

> 

这是我运行出来的错误

你把训练注释掉之前是否已经训练好模型了，然后也正确加载了模型。

from pytorch_bert_bilstm_crf_ner.

不知道我理解的对不对，您是指最后在final_data里生成pkl文件吗，这个我已经生成。程序中的read_pkl也能够读取。训练的模型不应该是经过bertForNer.train()才能生成吗？

from pytorch_bert_bilstm_crf_ner.

不知道我理解的对不对，您是指最后在final_data里生成pkl文件吗，这个我已经生成。程序中的read_pkl也能够读取。训练的模型不应该是经过bertForNer.train()才能生成吗？

可以预先处理，也可以在训练中生成，先生成pkl是为了避免每次都要重新处理耗时

from pytorch_bert_bilstm_crf_ner.

十分感谢，您说办法很有效，batch_size的大小对于能否成功运行很重要

from pytorch_bert_bilstm_crf_ner.