lyhue1991/torchkeras

Qwen-7b微调内容有误

evanguard opened this issue · 1 comments

evanguard commented

5分钟对话微调修改Qwen-7b的自我认知 这个错了吧， 咋是
from torchkeras.chat import ChatLLM
llm = ChatLLM(model,tokenizer)

https://github.com/lyhue1991/torchkeras/blob/master/examples/Qwen7B_MultiRounds%E2%80%94%E2%80%94transformers.ipynb

evanguard commented

看错了 ChatLLM 是封装吧 我先试下哈