Fine tuning blip2 with PPO

#34
by ksooklall - opened

I am following this notebook
https://github.com/huggingface/notebooks/blob/main/peft/Fine_tune_BLIP2_on_an_image_captioning_dataset_PEFT.ipynb

And I want to use the PPO trainer (https://huggingface.co./docs/trl/v0.8.6/en/ppo_trainer#ppo-trainer) to further fine-tune the model. Does anyone have experience with this, I can't find anything on google.

Thanks

Sign up or log in to comment