Reward Modeling on MR2Bench Video
50.7BoN ScoreMolmo2-4B RM
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| Molmo2-4B RMSize=4B2026.04 | 50.7 | 77.4 | 0.55 | |
| Qwen3-VL-32BSize=32B2026.04 | 49.9 | 77.8 | 0.592 | |
| InternVL3-78BSize=78B2026.04 | 49.5 | 76.7 | 0.553 | |
| IXC-2.5-RewardSize=7B2026.04 | 48.7 | 74 | 0.491 | |
| Qwen3-VL-4BSize=4B2026.04 | 47.9 | 75.8 | 0.538 | |
| Qwen3-VL-8BSize=8B2026.04 | 47.7 | 77.2 | 0.563 | |
| Qwen3-VL-4B RMSize=4B2026.04 | 47.5 | 73.8 | 0.482 | |
| Skywork-VL-RewardSize=7B2026.04 | 46.7 | 74.4 | 0.499 | |
| R1-RewardSize=7B2026.04 | 44.8 | 68.4 | 0.496 | |
| Qwen2.5-VL-7BSize=7B2026.04 | 44.4 | 73.6 | 0.504 | |
| Molmo2-4BSize=4B2026.04 | 43.2 | 71.9 | 0.485 | |
| Molmo2-8BSize=8B2026.04 | 42.6 | 70.8 | 0.476 | |
| InternVL3-8BSize=8B2026.04 | 40.4 | 71.6 | 0.466 | |
| LLaVA-CriticSize=7B2026.04 | 40.2 | 69 | 0.421 | |
| MM-RLHF-RewardSize=7B2026.04 | 36.6 | 67 | 0.352 |