Does `get_peft_model` return the frozen model

Hello, I have a question regarding doing the SFT with peft.

I’m thinking of using Lora, but not using SFTTrainer but with my trainer code.

But I’m not sure if I apply get_peft_model, I would get the model whose original part is already frozen, and only the adapter part is trainable.

Can anybody let me know this?

Thank you

1 Like