train_mem.py

# Adopted from tatsu-lab@stanford_alpaca. Below is the original copyright:
# Make it more memory efficient by monkey patching the LLaMA model with FlashAttn.

# Need to call this before importing transformers.
from fastchat.train.llama_flash_attn_monkey_patch import replace_llama_attn_with_flash_attn

replace_llama_attn_with_flash_attn()

from fastchat.train.train import train

if __name__ == "__main__":
    train()