Visual Reasoning on Geometry3K (test)
56.18Pass@1 AccuracySaEI
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| SaEIFinetuning status=Finetuned, Group size=n1 = n2 = 6 (total n = 12)2025.12 | 56.18 | — | |
| KL-CovFinetuning status=Finetuned, Group size=n = 122025.12 | 55.91 | — | |
| NoisyRolloutFinetuning status=Finetuned, Group size=n1 = n2 = 6 (total n = 12)2025.12 | 54.74 | — | |
| Vanilla GRPOFinetuning status=Finetuned, Group size=n = 122025.12 | 54.02 | — | |
| Qwen2.5-VL-7B-InstructFinetuning status=Not finetuned2025.12 | 39.27 | — | |
| GRPOBackbone=Qwen2.5-VL-3B-Instruct, Training Data=ViRL39K2026.03 | — | 28.72 | |
| PAPO_GBackbone=Qwen2.5-VL-3B-Instruct, Training Data=ViRL39K2026.03 | — | 30.95 | |
| PEPO_GBackbone=Qwen2.5-VL-3B-Instruct, Training Data=ViRL39K2026.03 | — | 29.85 |