MOSS-RLHF icon indicating copy to clipboard operation
MOSS-RLHF copied to clipboard

reward_model准确率

Open mingrenbuke opened this issue 1 year ago • 1 comments

想请教下开源的中英文reward_model的准确率大概是多少呢?

mingrenbuke avatar Jul 18 '23 10:07 mingrenbuke

您好,详见技术报告第十页,有中英文reward model在trainset 和 evalset上面的准确率

Ablustrund avatar Jul 19 '23 02:07 Ablustrund