Open ramsey-coding opened 1 year ago
I would like to get token embedding for python and java code.
I would be curious to know what the authors think about this?
Will the tokens for python code tokens will be meaning full?
Because CodeBERT and UniXcoder are pretrained on the MLM (Masked Language Model) objective, they are meaning full to some extent. Specifically, it still depends on the tasks you need to accomplish.
I would like to get token embedding for python and java code.
I would be curious to know what the authors think about this?
Will the tokens for python code tokens will be meaning full?