Question Answering on TriviaQA (EM, F1, Avg. iter., AUROC)
40EMAB-RAG
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| AB-RAGBackbone=Claude Haiku2026.06 | 40 | 53.5 | 1.81 | 0.733 | |
| Static RAGBackbone=Claude Haiku2026.06 | 35.5 | 50.8 | — | — |
| Method | Links | ||||
|---|---|---|---|---|---|
| AB-RAGBackbone=Claude Haiku2026.06 | 40 | 53.5 | 1.81 | 0.733 | |
| Static RAGBackbone=Claude Haiku2026.06 | 35.5 | 50.8 | — | — |