The parameter for inference

[yilun-lee]

Thank you for your excellent work on ZeroEval!

We are looking to integrate ZeroEval into our evaluation pipeline and would like to reproduce the results from the leaderboard. Could you please share the inference parameters used, such as temperature, top_p, and rp? Or are there any other specific settings we should adjust to match the leaderboard results?

Thank you for your help!