Multimodal Reasoning on SciQA
93.7AccuracyOpenMMReasoner-7B
Evaluation Results
| Method | Links | |
|---|---|---|
| OpenMMReasoner-7BEvaluation Source=reproduced2026.05 | 93.7 | |
| AnE-2ndTraining Stage=Round 22026.05 | 93.6 | |
| AnE-1stTraining Stage=Round 12026.05 | 93.5 | |
| AnE-3rdTraining Stage=Round 32026.05 | 93.5 | |
| Metis-RISEEvaluation Source=reproduced2026.05 | 93.4 | |
| MMR1Evaluation Source=reproduced2026.05 | 92.7 | |
| Preliminary RLTraining Phase=Warm-up2026.05 | 92.6 | |
| Revisual-R1Evaluation Source=reproduced2026.05 | 91.9 | |
| Vision-R1Evaluation Source=reproduced2026.05 | 91.4 | |
| SPARK-VL-7BEvaluation Source=original paper2026.05 | 90.8 | |
| VL-RethinkerEvaluation Source=original paper2026.05 | 89.6 | |
| OpenVLThinkerEvaluation Source=reproduced2026.05 | 88.9 | |
| Qwen2.5-VL-7B-InstructEvaluation Source=original paper2026.05 | 85.4 | |
| LLaVA-Critic-R1Evaluation Source=reproduced2026.05 | 84.2 |