Visual Math Reasoning on DynaMath (Pass@1)
79Pass@1SFT + AXPO
Evaluation Results
| Method | Links | |
|---|---|---|
| SFT + AXPOBase Model=Qwen3-VL-8B-Thinking2026.05 | 79 | |
| Qwen3-VL-8B-Thinking (Agent)Base Model=Qwen3-VL-8B-Thinking2026.05 | 75.9 | |
| PyVision-RLBase Model=Qwen2.5-VL-7B2026.05 | 61.6 | |
| DeepEyes-v2Base Model=Qwen2.5-VL-7B2026.05 | 57.2 | |
| DeepEyesBase Model=Qwen2.5-VL-7B2026.05 | 55 | |
| Qwen2.5-VL-7B-InstructBase Model=Qwen2.5-VL-7B2026.05 | 53.3 |