xfactlab / orpo

Official repository for ORPO
Apache License 2.0
376 stars 34 forks source link

Could you please tell me which OpenAI API you used during the MT-Bench evaluation? #34

Open hitszxs opened 2 weeks ago

hitszxs commented 2 weeks ago

Could you please tell me which OpenAI API you used during the MT-Bench evaluation? gpt-3.5 turbo or others?

jiwooya1000 commented 2 weeks ago

Hello @hitszxs, We used the default setting of the FastChat repo when we worked on this project, which makes GPT-4 an evaluator.

hitszxs commented 2 weeks ago

OK! Thank you for your response.

Hello @hitszxs, 你好 , We used the default setting of the FastChat repo when we worked on this project, which makes GPT-4 an evaluator. 我们在处理此项目时使用了 FastChat 存储库的默认设置,这使得 GPT-4 成为评估器。