模型是怎么调出来的

FreedomIntelligence / HuatuoGPT

HuatuoGPT, Towards Taming Language Models To Be a Doctor. (An Open Medical GPT)

Apache License 2.0

1.08k stars 142 forks source link

Closed bgtii closed 9 months ago

bgtii commented 9 months ago

用LLama-factory框架全参微调基座模型后，完全丧失了通用能力

StevenZHB commented 9 months ago

Hi @bgtii : 你好，HuatuoGPT 在训练时是混合了通用对话数据进行训练的，具体您可以参考论文中的设置。

Best, Hongbo