不同电脑运行出现IndexError: tensors used as indices must be long, byte or bool tensors问题

[okfmeplz]

换一个电脑之后运行main.py报错：
Traceback (most recent call last):
File "G:/BERT-Relation-Extraction/BERT-Relation-Extraction/ner_main.py", line 190, in
main(data_name)
File "G:/BERT-Relation-Extraction/BERT-Relation-Extraction/ner_main.py", line 182, in main
train.train()
File "G:/BERT-Relation-Extraction/BERT-Relation-Extraction/ner_main.py", line 52, in train
output = self.model(input_ids, attention_mask, labels)
File "C:\ProgramData\Anaconda3\envs\pytorchpy38_gpu_cuda111\lib\site-packages\torch\nn\modules\module.py", line 1051, in _call_impl
return forward_call(*input, **kwargs)
File "G:\BERT-Relation-Extraction\BERT-Relation-Extraction\model.py", line 37, in forward
loss = -self.crf(seq_out, labels, mask=attention_mask.bool(), reduction='mean')
File "C:\ProgramData\Anaconda3\envs\pytorchpy38_gpu_cuda111\lib\site-packages\torch\nn\modules\module.py", line 1051, in call_impl
return forward_call(*input, **kwargs)
File "C:\ProgramData\Anaconda3\envs\pytorchpy38_gpu_cuda111\lib\site-packages\torchcrf_init.py", line 102, in forward
numerator = self.compute_score(emissions, tags, mask)
File "C:\ProgramData\Anaconda3\envs\pytorchpy38_gpu_cuda111\lib\site-packages\torchcrf_init.py", line 186, in _compute_score
score = self.start_transitions[tags[0]]
IndexError: tensors used as indices must be long, byte or bool tensors
请问大佬知道原因吗？ TAT

[taishan1994]

可能是torch版本的问题。

[okfmeplz]

if labels is not None:
  labels=labels.long()
  loss = -self.crf(seq_out, labels, mask=attention_mask.bool(), reduction='mean')

已解决，在model里错误代码前一句添加 labels=labels.long()