List your company

Multimodal Reasoning on MMMU (RC-Acc)

57.8RC-AccGroupwise Ranking Reward
47.81650.4085355.592Apr 20, 2026

Evaluation results

MethodLinks
Groupwise Ranking RewardReward Method=Groupwise Ranking Reward, Training Budget=Matched, Training Epochs=22026.0457.8
Pointwise GRReward Method=Pointwise GR, Training Budget=Matched, Training Epochs=22026.0456.6
Qwen2.5-VL-7B-ITReference Checkpoint=Qwen2.5-VL-7B-IT2026.0452.6
PRMReward Method=PRM, Training Budget=Matched, Training Epochs=22026.0450
RLVRReference Checkpoint=RLVR2026.0448.2