Question Answering on HotpotQA distractor (Clean)
69AccuracyAdaptive-k
Evaluation results
| Method | Links | ||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|
| Adaptive-kk=3, N=200, Evaluator=GPT-4o2026.05 | 69 | 27.8 | 82 | 40.8 | 1,116 | 8.6 | 1,118 | 67 | 63 | 59 | |
| SEAL-RAGk=3, N=200, L=3, Evaluator=GPT-4o2026.05 | 68.5 | 80.8 | 42 | 54.9 | 136 | 1 | 134 | 26 | 23 | 23 | |
| ADAGATEk=3, N=200, L=3, Evaluator=GPT-4o2026.05 | 64.5 | 54.9 | 74.5 | 62 | 370 | 3 | 347 | 54 | 50 | 48 | |
| ADAGATEk=3, N=200, L=1, Evaluator=GPT-4o2026.05 | 63 | 55.5 | 74 | 62.3 | 360 | 2.9 | 338 | 58 | 51 | 51 | |
| SEAL-RAGk=3, N=200, L=1, Evaluator=GPT-4o2026.05 | 62 | 78.4 | 42 | 54.1 | 140 | 1.1 | 139 | 24 | 22 | 22 | |
| Self-RAGk=3, N=200, Evaluator=GPT-4o2026.05 | 60.5 | 64.5 | 50 | 53.6 | 234 | 1.9 | 254 | 47 | 44 | 44 | |
| Basic RAGk=3, N=200, Evaluator=GPT-4o2026.05 | 58.5 | 55.9 | 66.5 | 60.1 | 364 | 3 | 352 | 50 | 44 | 44 |