greyweb opened this issue a year ago · 0 comments
Hi, I am trying to finetune GPT-J 6B from HF converted weights. It would be great to know some recommendations on the finetuning compute widely used/ suggested for GPT-J 6B.