Perception on HRBen 4K
91Pass@4AXPO
Evaluation Results
| Method | Links | |
|---|---|---|
| AXPOModel Size=4B, Backbone=Qwen3-VL-Thinking2026.05 | 91 | |
| SFTModel Size=8B, Backbone=Qwen3-VL-Thinking2026.05 | 90.5 | |
| BaseModel Size=32B, Backbone=Qwen3-VL-Thinking2026.05 | 90.5 | |
| AXPOModel Size=8B, Backbone=Qwen3-VL-Thinking2026.05 | 90 | |
| SFTModel Size=4B, Backbone=Qwen3-VL-Thinking2026.05 | 89.5 | |
| AXPOModel Size=2B, Backbone=Qwen3-VL-Thinking2026.05 | 88.5 | |
| SFT + GRPOModel Size=4B, Backbone=Qwen3-VL-Thinking2026.05 | 88 | |
| SFT + GRPOModel Size=8B, Backbone=Qwen3-VL-Thinking2026.05 | 87.5 | |
| GRPOModel Size=4B, Backbone=Qwen3-VL-Thinking2026.05 | 87 | |
| GRPOModel Size=8B, Backbone=Qwen3-VL-Thinking2026.05 | 87 | |
| SFTModel Size=2B, Backbone=Qwen3-VL-Thinking2026.05 | 86.5 | |
| BaseModel Size=8B, Backbone=Qwen3-VL-Thinking2026.05 | 85.5 | |
| SFT + GRPOModel Size=2B, Backbone=Qwen3-VL-Thinking2026.05 | 83.5 | |
| BaseModel Size=4B, Backbone=Qwen3-VL-Thinking2026.05 | 83.5 | |
| BaseModel Size=2B, Backbone=Qwen3-VL-Thinking2026.05 | 79 | |
| GRPOModel Size=2B, Backbone=Qwen3-VL-Thinking2026.05 | 75.5 |