trlx icon indicating copy to clipboard operation
trlx copied to clipboard

Support for CodeGenForCausalLM

Open ZhengYang opened this issue 1 year ago • 3 comments

🚀 The feature, motivation, and pitch

I'm trying to apply RL in a code generation LM: https://huggingface.co/docs/transformers/model_doc/codegen, unfortunately getting the error below:

ValueError: Unsupported architecture: `CodeGenForCausalLM`. The following architectures are available for model 
branching:
['GPTJForCausalLM', 'GPT2LMHeadModel', 'GPTNeoForCausalLM', 'GPTNeoXForCausalLM', 'OPTForCausalLM', 'BloomModel', 
'BloomForCausalLM', 'LlamaModel', 'LlamaForCausalLM']

I wonder if it would be possible to add support for code modes. Thanks!

Alternatives

No response

Additional context

No response

ZhengYang avatar Jun 03 '23 15:06 ZhengYang

@jon-tow

LouisCastricato avatar Jun 03 '23 19:06 LouisCastricato

mark

huybery avatar Jun 05 '23 09:06 huybery

Also there is neeed of Support for GPTBigCodeForCausalLM

22Mukesh22 avatar Jun 06 '23 06:06 22Mukesh22