Visual Question Answering on GEOQA-8K
62.04Average ScoreReSum
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| ReSumBackbone=Qwen2.5-3B-VL-Instruct2026.06 | 62.04 | 5.61 | |
| DGPOBackbone=Qwen2.5-3B-VL-Instruct2026.06 | 59.95 | 2.52 | |
| GPGBackbone=Qwen2.5-3B-VL-Instruct2026.06 | 59.02 | 1.59 | |
| DAPOBackbone=Qwen2.5-3B-VL-Instruct2026.06 | 59.02 | 1.59 | |
| GRPO-ADBackbone=Qwen2.5-3B-VL-Instruct2026.06 | 58.09 | 0.66 | |
| Dr. GRPOBackbone=Qwen2.5-3B-VL-Instruct2026.06 | 57.96 | 0.53 | |
| GRPOBackbone=Qwen2.5-3B-VL-Instruct2026.06 | 57.43 | — | |
| GSPOBackbone=Qwen2.5-3B-VL-Instruct2026.06 | 57.16 | -0.27 | |
| Base ModelBackbone=Qwen2.5-3B-VL-Instruct2026.06 | 39.79 | — |