Save bert-base-uncased model as checkpoint

🤗Transformers
1

I am trying to save bert-base-uncased model as checkpoint with torch.save{ model.state_dict()} when i load the model to resume training i get
typeError: ‘BertForTokenClassification’ object is not subscriptable.
I want to save model as checkpoint and load it with resume model training. any help??
it’s notebook used for train my model “Google Colab”