Fine-tune code on 1b Model

[kokolerk]

hi @kazemnejad
What a wonderful idea and insight!
NOW I want to search more work after you, first I want to fine-tune your nope model. But I can only see the inference code in the readme. Could you please tell me how to fine-tune on your model?

[kazemnejad]

Thanks for reaching out.

These are standard Huggingface models, which can be plugged into finetuning pipeline right away.

https://huggingface.co/docs/trl/sft_trainer

[kokolerk]

Thanks for your kind reply. That helps a lot.

Also I want to ask one more question, could you show me your pertaining dataset? Because I don't want to make distribution shift, so maybe the same source data for fine-tune is the best way.

Thank you again for your help.