PPO data en

[borisshapa]

Hello!

I really liked your work on RLHF research. A very clear description in the technical report and a good implementation. I studied the entire code in detail and read the article several times.

For my research, I would like to reproduce your results, but I can't find a dataset with English prompts that was used in the PPO algorithm. In the article you write that a manually collected dataset was used, but I can't find it anywhere. Could you share this dataset so I can run your code, please?

Thanks

[Ablustrund]

Hi, thank you for your interest in this work!

We have cleaned the dataset with some filtering algorithms, but for some confidentiality reasons, we can't release the dataset at the moment.

sorry about that