已解决：KeyError ‘attention_mask‘

已解决：KeyError: ‘attention_mask’

调用bert模型预训练时报错KeyError: ‘attention_mask’

已知是transformers与tokenizers的版本号不匹配，导致attention_mask丢失

可使用 pip show 包查看对应版本号

Name: transformers
Version: 2.8.0
Summary: State-of-the-art Natural Language Processing for TensorFlow 2.0 and PyTorch
Home-page: https://github.com/huggingface/transformers
Author: Thomas Wolf, Lysandre Debut, Victor Sanh, Julien Chaumond, Google AI Language Team Authors, Open AI team Authors, Facebook AI Authors, Carnegie Mellon University Authors
Author-email: thomas@huggingface.co
License: Apache
Location: c:\programdata\anaconda3\envs\win_torch\lib\site-packages
Requires: sacremoses, regex, requests, tqdm, numpy, sentencepiece, boto3
Required-by:

Name: tokenizers
Version: 0.5.2
Summary: Fast and Customizable Tokenizers
Home-page: https://github.com/huggingface/tokenizers
Author: Anthony MOI
Author-email: anthony@huggingface.co
License: Apache License 2.0
Location: c:\programdata\anaconda3\envs\win_torch\lib\site-packages
Requires:
Required-by: transformers

目前已知 transformers=2.8.0，tokenizers=0.5.2，或 transformers=3.4.0，tokenizers=0.9.2

可以解决此问题