Object Recognition on VGGFace2 (test)
100Original For.GA
Evaluation Results
| Method | Links | |||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| GABase Model=Qwen2.5-VL-3B-Instruct2026.05 | 100 | 0 | 91.67 | 12.86 | 5.33 | 97.29 | 51.19 | 0 | 59.87 | |
| GDBase Model=Qwen2.5-VL-3B-Instruct2026.05 | 100 | 0 | 87.33 | 17.57 | 23 | 74.14 | 50.34 | 0 | 58.53 | |
| Qwen2.5-VL-3B-Instruct + Stage 1Stage=Stage 12026.05 | 99.67 | 90.71 | 99.33 | 87.43 | 91.67 | 75.14 | 90.66 | 86.67 | 60.5 | |
| Qwen2.5-VL-3B-Instruct + Stage 2Stage=Stage 22026.05 | 99.67 | 99.43 | 99.33 | 99.86 | 50.33 | 99.86 | 91.41 | 0 | 60.34 | |
| HFRUBase Model=Qwen2.5-VL-3B-Instruct, Stage=Final2026.05 | 99.67 | 99.71 | 99.67 | 99.57 | 96.33 | 99.29 | 99.04 | 1.34 | 60.47 | |
| UNDIALBase Model=Qwen2.5-VL-3B-Instruct2026.05 | 99 | 8.43 | 99 | 18.29 | 8.67 | 97.29 | 55.11 | 1 | 59.92 | |
| NPOBase Model=Qwen2.5-VL-3B-Instruct2026.05 | 98.67 | 3.71 | 38.67 | 52.14 | 1.33 | 98.86 | 48.9 | 0 | 60.09 | |
| SLUGBase Model=Qwen2.5-VL-3B-Instruct2026.05 | 95.67 | 22.29 | 76.33 | 49.57 | 12.33 | 94.86 | 58.51 | 4.67 | 26.28 | |
| SimNPOBase Model=Qwen2.5-VL-3B-Instruct2026.05 | 95 | 4.57 | 99.67 | 0.29 | 35.33 | 83.57 | 53.07 | 100 | 59.74 | |
| SAUCEBase Model=Qwen2.5-VL-3B-Instruct2026.05 | 93.67 | 70.57 | 94.33 | 61.14 | 92 | 70.57 | 80.38 | 20.67 | 49.68 | |
| RMUBase Model=Qwen2.5-VL-3B-Instruct2026.05 | 69.67 | 46.14 | 92 | 15.71 | 41.67 | 96.71 | 60.32 | 0.67 | 59.05 | |
| Qwen2.5-VL-3B-InstructStatus=Base Model2026.05 | 22.33 | 80.57 | 17.67 | 81.14 | 1.33 | 99.29 | 50.39 | 3.67 | 59.92 |