Localisation on PRISM 1.0 (val)
78.33Readability IoUQwenExpert
Evaluation Results
| Method | Links | ||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| QwenExpertModel type=Fine-tuned, Training data=PRISM annotations2026.05 | 78.33 | 80.15 | 81.32 | 79.98 | 67.61 | 70.24 | 74.52 | 71.96 | 73.28 | 75.53 | 79.14 | 77.3 | |
| GPT-4o2026.05 | 55.32 | 64.33 | 62.28 | 60.37 | 51.64 | 58.32 | 62.15 | 60.17 | 54.17 | 61.12 | 63.54 | 62.1 | |
| GPT-4o-mini2026.05 | 40.28 | 43.87 | 61.27 | 47.18 | 34.56 | 39.24 | 53.11 | 44.78 | 37.29 | 40.12 | 52.38 | 45.26 | |
| QwenBaseModel type=Base2026.05 | 36.45 | 40.91 | 54.78 | 43.12 | 29.52 | 33.14 | 47.25 | 38.04 | 32.15 | 34.11 | 45.13 | 37.7 | |
| QwenPromptedModel type=Prompted2026.05 | 36.03 | 43.03 | 47.82 | 41.47 | 31.86 | 35.19 | 48.84 | 40.49 | 32.97 | 35.2 | 47.28 | 39.84 |