Spaces:

allenai

/

reward-bench

like 275

Running

@0-hero we'll be faster to update if you open an issue on github, and clearly give us pairs of dpo-refmodel pairs. Regardless, we'll do it soon probably.

Thanks Nat, will open on Github next time.

DPO models are

1. 0-hero/Matter-0.1-7B-DPO-preview
2. 0-hero/Matter-0.1-7B-boost-DPO-preview

Results are added to the dataset, will show up in leaderboard at next restart.

Thanks Nat! :)