Multimodal Reasoning on M3CoT
82.45NS ScorePRCR
Evaluation Results
| Method | Links | ||||||
|---|---|---|---|---|---|---|---|
| PRCRModel=Qwen3-VL, #Params=32B2026.06 | 82.45 | 68.1 | 96.44 | 95.53 | 45.8 | 76.86 | |
| Token-ReplayModel=Qwen3-VL, #Params=32B2026.06 | 82.12 | 67.52 | 96.21 | 95.12 | 45.23 | 76.49 | |
| BaselineModel=Qwen3-VL, #Params=32B2026.06 | 80.59 | 65.76 | 96.21 | 94.31 | 43.57 | 75.28 | |
| PRCRModel=Qwen3-VL, #Params=8B2026.06 | 75.86 | 60.19 | 88.63 | 92.68 | 24.07 | 68.68 | |
| Token-ReplayModel=Qwen3-VL, #Params=8B2026.06 | 75.48 | 59.71 | 89.1 | 91.87 | 24.9 | 68.46 | |
| BaselineModel=Qwen3-VL, #Params=8B2026.06 | 73.05 | 56.37 | 88.15 | 93.5 | 21.99 | 66.44 | |
| PRCRModel=InternVL3.5, #Params=8B2026.06 | 68.45 | 52.71 | 91 | 86.99 | 34.85 | 65.4 | |
| Token-ReplayModel=InternVL3.5, #Params=8B2026.06 | 67.94 | 49.36 | 91.94 | 90.24 | 34.44 | 64.75 | |
| PRCRModel=InternVL3.5, #Params=14B2026.06 | 67.84 | 49.85 | 85.26 | 88.62 | 49.75 | 65.93 | |
| Token-ReplayModel=InternVL3.5, #Params=14B2026.06 | 67.28 | 49.18 | 85.5 | 88.21 | 49.06 | 65.5 | |
| BaselineModel=InternVL3.5, #Params=8B2026.06 | 67.05 | 51.27 | 90.52 | 85.37 | 31.95 | 64.06 | |
| BaselineModel=InternVL3.5, #Params=14B2026.06 | 66.16 | 47.93 | 84.83 | 86.99 | 47.72 | 64.28 |