Mathematical Reasoning on MathVista (ASR, Coh, Rat, Acc)
77.9AccuracyQwen3-VL-4B-Instruct / H-OPD (8B)
Evaluation Results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| Qwen3-VL-4B-Instruct / H-OPD (8B)Student=Qwen3-VL-4B-Instruct, Teacher=Qwen3-VL-8B + Qwen3-8B, Method=H-OPD2026.07 | 77.9 | — | — | — | — | |
| Qwen3-VL-8BType=Baseline2026.07 | 77.2 | — | — | — | — | |
| Qwen3-VL-4B-Instruct / ExOPD (8B)Student=Qwen3-VL-4B-Instruct, Teacher=Qwen3-VL-8B, Method=ExOPD2026.07 | 77.2 | — | — | — | — | |
| Qwen3-VL-4B-Instruct / OPD (8B)Student=Qwen3-VL-4B-Instruct, Teacher=Qwen3-VL-8B, Method=OPD2026.07 | 76.6 | — | — | — | — | |
| Qwen3-VL-4B-Instruct / GRPOStudent=Qwen3-VL-4B-Instruct, Teacher=Self, Method=GRPO2026.07 | 76.1 | — | — | — | — | |
| Qwen3-VL-4BType=Baseline2026.07 | 73.7 | — | — | — | — | |
| Qwen3-VL-4B-Instruct / SFT (8B)Student=Qwen3-VL-4B-Instruct, Teacher=Qwen3-VL-8B, Method=SFT2026.07 | 72.1 | — | — | — | — | |
| Qwen3-VL-2B-Instruct / H-OPD (8B)Student=Qwen3-VL-2B-Instruct, Teacher=Qwen3-VL-8B + Qwen3-8B, Method=H-OPD2026.07 | 67.2 | — | — | — | — | |
| Qwen3-VL-2B-Instruct / H-OPD (4B)Student=Qwen3-VL-2B-Instruct, Teacher=Qwen3-VL-4B + Qwen3-4B, Method=H-OPD2026.07 | 66.8 | — | — | — | — | |
| Qwen3-VL-2B-Instruct / ExOPD (8B)Student=Qwen3-VL-2B-Instruct, Teacher=Qwen3-VL-8B, Method=ExOPD2026.07 | 66.1 | — | — | — | — | |
| Qwen3-VL-2B-Instruct / OPD (8B)Student=Qwen3-VL-2B-Instruct, Teacher=Qwen3-VL-8B, Method=OPD2026.07 | 65.7 | — | — | — | — | |
| Qwen3-VL-2B-Instruct / ExOPD (4B)Student=Qwen3-VL-2B-Instruct, Teacher=Qwen3-VL-4B, Method=ExOPD2026.07 | 64.4 | — | — | — | — | |
| Qwen3-VL-2B-Instruct / OPD (4B)Student=Qwen3-VL-2B-Instruct, Teacher=Qwen3-VL-4B, Method=OPD2026.07 | 64.2 | — | — | — | — | |
| Qwen3-VL-2B-Instruct / GRPOStudent=Qwen3-VL-2B-Instruct, Teacher=Self, Method=GRPO2026.07 | 63.6 | — | — | — | — | |
| Qwen3-VL-2B-Instruct / SFT (8B)Student=Qwen3-VL-2B-Instruct, Teacher=Qwen3-VL-8B, Method=SFT2026.07 | 62.2 | — | — | — | — | |
| Qwen3-VL-2B-Instruct / SFT (4B)Student=Qwen3-VL-2B-Instruct, Teacher=Qwen3-VL-4B, Method=SFT2026.07 | 61.6 | — | — | — | — | |
| Qwen3-VL-2BType=Baseline2026.07 | 61.3 | — | — | — | — | |
| BaseBase model=Qwen2.5-VL-7B2026.07 | 49 | — | 3.42 | 3.19 | — | |
| ReShiftBase model=Qwen2.5-VL-7B2026.07 | 49 | 85 | 3.02 | 2.99 | 7 | |
| RewriteBase model=Qwen2.5-VL-7B2026.07 | 47 | 74 | 2.68 | 2.92 | 24 | |
| BadVisionBase model=Qwen2.5-VL-7B2026.07 | 46 | 30 | 2.67 | 2.79 | 13 | |
| BadTokenBase model=Qwen2.5-VL-7B2026.07 | 45 | 77 | 2.86 | 2.72 | 22 |