embedding training config file

[starry-y]

Thanks for your work!

I can not find this file train-embeddings-base-1gpu.json mentioned in ReadMe.md, but found bert-wwm-ext_literature file. Does the bert-wwm-ext_literature file replace the former file?

Thanks a lot!

[Vimos]

Hi, the basic difference of configurations are db paths. For embeddings, we use literature data rather than official data as the training data.

Yes, please use the ext_literature as the configuration file.

[starry-y]

Ok, thanks for your reply. I have replace the config file in the terminal.

And I have another question.

In the evaluation stage, what is pretrained/Chinese-word-vector/embeddings refering to ？

[starry-y]

And I could not find chengyu_synonym_dict in train_embedding.py ...

Sorry for bothering you, and waiting for your reply.

[Vimos]

Please refer to https://github.com/VisualJoyce/ChengyuBERT#learning-and-evaluating-chinese-idiom-embeddings

This is a different paper focusing on embedding learning and evaluation. The data has been shared online.