List your company

Humor on HH-RLHF (test)

2.481RewardPPO
2.357242.389372.42152.45363Aug 19, 2025

Evaluation results

MethodLinks
PPOModel Size=13B2025.082.48112.14
MAVISModel Size=13B2025.082.4659.2
MAVISModel Size=7B2025.082.3763.78
PPOModel Size=7B2025.082.36210.43