train_xformers.py 362 Bytes
Newer Older
dongchy920's avatar
dongchy920 committed
1
2
3
4
5
6
7
8
9
10
11
12
13
# Make it more memory efficient by monkey patching the LLaMA model with xformers attention.

# Need to call this before importing transformers.
from mgm.train.llama_xformers_attn_monkey_patch import (
    replace_llama_attn_with_xformers_attn,
)

replace_llama_attn_with_xformers_attn()

from mgm.train.train import train

if __name__ == "__main__":
    train()