Image Captioning on Prism Framework
80.5Chart QA AccuracyCapRL++
Evaluation results
| Method | Links | ||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| CapRL++GRPO Trained=true2026.06 | 80.5 | 39.9 | 64.8 | 30.7 | 36.4 | 32.4 | 60.1 | 23.4 | 55 | 70.6 | 38.1 | 48.3 | |
| Qwen2.5-VL-72BGRPO Trained=false2026.06 | 80.2 | 38 | 60.8 | 34.1 | 39.9 | 30.7 | 60.2 | 24.5 | 55 | 69.3 | 39.4 | 48.3 | |
| Qwen2.5-VL-7BGRPO Trained=false2026.06 | 74.9 | 35.4 | 56.4 | 30.1 | 36.4 | 24.8 | 57 | 23.3 | 50.7 | 67.1 | 37.9 | 44.9 | |
| Qwen2.5-VL-as-Judge-3BGRPO Trained=true2026.06 | 71.4 | 34.2 | 49.3 | 29.1 | 33.8 | 22.9 | 54.3 | 24.1 | 47.7 | 64.5 | 36.4 | 42.5 | |
| Qwen2.5-VL-3BGRPO Trained=false2026.06 | 65.6 | 27.1 | 40.2 | 28.6 | 32.8 | 21.8 | 54.4 | 22.6 | 46.4 | 64.1 | 35.1 | 39.9 | |
| UnifiedRW-as-Judge-3BGRPO Trained=true2026.06 | 54.9 | 25.1 | 33.6 | 28.1 | 34.6 | 20.4 | 58.2 | 24.5 | 45.4 | 61.2 | 36.3 | 38.4 |