search

huggingface / huggingface-llama-recipes

531 stars 59 forks source link

Add SFT recipe #36

Closed lewtun closed 1 month ago

lewtun commented 1 month ago

Ports the SFT recipe for the Llama Vision models from TRL.

lewtun commented 1 month ago

Very nice! I believe we're trying to have notebooks only in this repo (but we're not doing a super great job at it 😁)

Ah makes sense. I'll merge for now and see if it's possible to get these big models running in a notebook with QLoRA later