visual-arena / fastchat /train /train_xformers.py
tianleliphoebe's picture
Upload folder using huggingface_hub
ec0c335 verified
# Make it more memory efficient by monkey patching the LLaMA model with xformers attention.
# Need to call this before importing transformers.
from fastchat.train.llama_xformers_attn_monkey_patch import (
replace_llama_attn_with_xformers_attn,
)
replace_llama_attn_with_xformers_attn()
from fastchat.train.train import train
if __name__ == "__main__":
train()