RTL Area Prediction on DeepRTL2 1.0 (test)
0.805R2 ScoreDeepRTL2
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| DeepRTL2Backbone=DeepSeek, Hard Negatives=Excluded2025.05 | 0.805 | 0.0295 | 0.445 | |
| DeepRTL2Backbone=DeepSeek, Hard Negatives=Included2025.05 | 0.773 | 0.016 | 0.476 | |
| DeepRTL2Backbone=Llama, Hard Negatives=Included2025.05 | 0.759 | 0.0197 | 0.49 | |
| text-embedding-3-large2025.05 | 0.699 | 0.0445 | 0.548 | |
| GritLM-7B2025.05 | 0.651 | 0.0388 | 0.591 | |
| text-embedding-3-small2025.05 | 0.603 | 0.0557 | 0.63 | |
| DeepRTL2Backbone=Llama, Hard Negatives=Excluded2025.05 | 0.51 | 0.0283 | 0.7 |