Section Retrieval on TriviaQA 250 questions
30.4Suppress RateBaseline
Evaluation Results
| Method | Links | ||||||
|---|---|---|---|---|---|---|---|
| BaselineStage=Baseline, Model=Qwen 3.5 2B, Threshold=0.802026.04 | 30.4 | — | — | — | 0 | 0.9 | |
| Evolve (Warm)Stage=Warm, Model=Qwen 3.5 2B, Threshold=0.802026.04 | 82 | 82.8 | 81.2 | 28.8 | 0.96 | 10.3 | |
| Evolve (Cold)Stage=Cold, Model=Qwen 3.5 2B, Threshold=0.802026.04 | 82.4 | 83.4 | 81.6 | 0 | 1.26 | 11.4 | |
| Evolve (Post-consolidation)Stage=Post-consolidation, Model=Qwen 3.5 2B, Threshold=0.802026.04 | 83.2 | 84 | 82.4 | 29.2 | 0.96 | 8.5 |