search

baichuan-inc / Baichuan2

A series of large language models developed by Baichuan Intelligent Technology
https://huggingface.co/baichuan-inc
Apache License 2.0
4.03k stars 286 forks source link

对齐的框架和数据 #13

Open skepsun opened 10 months ago

skepsun commented 10 months ago

看了论文，baichuan2 chat版本做了rlhf流程，采集了类似于hh_rlhf的数据，请问有开源rlhf数据和训练框架的计划吗？或者可以先开源一部分reward model训练数据？

jieli4970 commented 10 months ago

附问一下，chat和base的差别就是加了对其这一步吗