-
# FastChat
An open platform for training, serving, and evaluating large language model based chatbots.
https://github.com/lm-sys/FastChat
# 설치방법
transformers 버전에 주의하자.
## 1. pip install
pip3 install fschat pip install git+https://github.com/huggingface/transformers@cae78c46d
# Vicuna Weights 얻기
1. LLaMA weights 획득 (혹은 https://huggingface.co/decapoda-research/llama-13b-hf 이용)
2. 다음 명령어 수행
python3 -m fastchat.model.apply_delta \ --base /path/to/llama-13b \ --target /output/path/to/vicuna-13b \ --delta lmsys/vicuna-13b-delta-v0
## 에러 해결
1. ValueError: Tokenizer class LLaMATokenizer does not exist or is not currently imported.
-> tokenizer_config.json를 "tokenizer_class": "LLaMATokeniers를 "tokenizer_class": "LlamaTokeniers로 변경
2. tokenizers>=0.13.3 is required for a normal functioning of this module, but found tokenizers==0.12.1.
-> transformers를 pip install git+https://github.com/huggingface/transformers@cae78c46d 설치