fine tune 65B model with A100 OOM
ChaoGaoUCR opened this issue · comments
Dear author, thanks for providing such an incredible project.
I am trying to fine-tune the 65B model with A100 on 40GB version,
However, it went OOM and I wondered is that was because the batch size was too big or is there was any way I could resolve this issue?
I have multiple GPUs available, and I wondered if there is any command I can use for parallel fine-tuning?
Thanks
Sorry I got it solved