how to train llama2-7b in a100 80G gpu?

[brewswang]

how to train llama2-7b in a100 80G gpu?

[young-geng]

It would be difficult to do that without some parameter efficient finetuning methods such as LoRA. For the full 7B model, the parameters and optimizer states would take 84GB of memory, exceeding the 80GB of a single A100.

[bet0x]

@young-geng Got a table or something similar with those requirements by model/size?

[kuangdao]

dose not support zero of jax？