List your company

Multimodal Reasoning on MMMU-Pro (RC-Acc)

39RC-AccPointwise GR
30.26432.53234.837.068Apr 20, 2026

Evaluation results

MethodLinks
Pointwise GRReward Method=Pointwise GR, Training Budget=Matched, Training Epochs=22026.0439
Groupwise Ranking RewardReward Method=Groupwise Ranking Reward, Training Budget=Matched, Training Epochs=22026.0438.8
RLVRReference Checkpoint=RLVR2026.0436.2
Qwen2.5-VL-7B-ITReference Checkpoint=Qwen2.5-VL-7B-IT2026.0435.2
PRMReward Method=PRM, Training Budget=Matched, Training Epochs=22026.0430.6