Is it possible to further fine tune from a checkpoint?

[thanhnew2001]

Hello
Thanks for the great repo. I just want to ask if it possible to further fine tune a checkpoint based on new dataset?
I saw this but not sure how it works:
https://github.com/mzbac/qlora-fine-tune/blob/main/merge_peft_adapters.py

Thanks
Thanh

[mzbac]

I'm not sure if I understand the question correctly, but you can fine-tune a fine-tuned model and the script is for merging the Lora adapter back to the base model so you can load the merged model as a normal transformer model.

[thanhnew2001]

Thank you for your email. I will try to merge and further fine tune the merged model with new dataset. Will let u know how it goes

On Sun, 16 Jul 2023 at 19:02 Anchen ***@***.***> wrote: I'm not sure if I understand the question correctly, but you can fine-tune a fine-tuned model and the script is for merging the Lora adapter back to the base model so you can load the merged model as a normal transformer model. — Reply to this email directly, view it on GitHub <#8 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AAY4GWEKQL3UWYP7TGZ74OLXQPJ4ZANCNFSM6AAAAAA2LY3CMU> . You are receiving this because you authored the thread.Message ID: ***@***.***>

-- Best regards, Thanh