Question Answering on HotpotQA distractor rho=0.5
64.5Accuracy (Acc)Adaptive-k
Evaluation results
| Method | Links | ||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|
| Adaptive-kk=3, N=200, Evaluator=GPT-4o2026.05 | 64.5 | 28.6 | 80 | 41.4 | 1,073 | 8.8 | 1,057 | 62 | 57 | 57 | |
| SEAL-RAGk=3, N=200, L=3, Evaluator=GPT-4o2026.05 | 61.5 | 80 | 41 | 54 | 131 | 1 | 127 | 21 | 19 | 19 | |
| Basic RAGk=3, N=200, Evaluator=GPT-4o2026.05 | 59.5 | 59.8 | 63 | 60.5 | 364 | 3 | 330 | 42 | 38 | 36 | |
| Self-RAGk=3, N=200, Evaluator=GPT-4o2026.05 | 59 | 65.5 | 48.5 | 53.4 | 211 | 1.7 | 240 | 37 | 32 | 32 | |
| ADAGATEk=3, N=200, L=3, Evaluator=GPT-4o2026.05 | 57.5 | 56.8 | 73.5 | 62.7 | 344 | 2.8 | 309 | 57 | 50 | 48 | |
| SEAL-RAGk=3, N=200, L=1, Evaluator=GPT-4o2026.05 | 57 | 79 | 40.8 | 53.5 | 131 | 1 | 126 | 20 | 18 | 17 | |
| ADAGATEk=3, N=200, L=1, Evaluator=GPT-4o2026.05 | 54 | 57.1 | 72.8 | 62.7 | 328 | 2.8 | 302 | 51 | 44 | 45 |