Search on MM Search (Pass@4)
61Pass@4AXPO
Evaluation Results
| Method | Links | |
|---|---|---|
| AXPOModel Size=8B, Backbone=Qwen3-VL-Thinking2026.05 | 61 | |
| GRPOModel Size=8B, Backbone=Qwen3-VL-Thinking2026.05 | 60 | |
| SFTModel Size=4B, Backbone=Qwen3-VL-Thinking2026.05 | 59.7 | |
| SFT + GRPOModel Size=8B, Backbone=Qwen3-VL-Thinking2026.05 | 59.3 | |
| BaseModel Size=32B, Backbone=Qwen3-VL-Thinking2026.05 | 59.3 | |
| BaseModel Size=8B, Backbone=Qwen3-VL-Thinking2026.05 | 59 | |
| AXPOModel Size=2B, Backbone=Qwen3-VL-Thinking2026.05 | 57.3 | |
| SFT + GRPOModel Size=4B, Backbone=Qwen3-VL-Thinking2026.05 | 57 | |
| AXPOModel Size=4B, Backbone=Qwen3-VL-Thinking2026.05 | 57 | |
| SFT + GRPOModel Size=2B, Backbone=Qwen3-VL-Thinking2026.05 | 55.7 | |
| BaseModel Size=4B, Backbone=Qwen3-VL-Thinking2026.05 | 55.7 | |
| SFTModel Size=8B, Backbone=Qwen3-VL-Thinking2026.05 | 55.3 | |
| GRPOModel Size=2B, Backbone=Qwen3-VL-Thinking2026.05 | 54.7 | |
| SFTModel Size=2B, Backbone=Qwen3-VL-Thinking2026.05 | 53.7 | |
| BaseModel Size=2B, Backbone=Qwen3-VL-Thinking2026.05 | 46.3 | |
| GRPOModel Size=4B, Backbone=Qwen3-VL-Thinking2026.05 | 42 |