Pretrain data

[xieexiaotuzi]

Hi, Thanks for sharing such a great work.

I would like to ask that do use remove the 'docsring' from the code of the codesearchnet dataseet when you pretrain the CodeBert? In detail, whether you use the 'code' for code generator and 'docstring' for nl generator; or use 'code(remove docstring from the code)' for code generator and 'docstring' for nl generator?

Thanks for your help.

[guoday]

'code(remove docstring from the code)' for code generator and 'docstring' for nl generator