Multi-hop Question Answering on Bamboogle (LJFT, LLMJ, ACC)
61.6LJFT ScoreSimpleDeepSearcher
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| SimpleDeepSearcherMode=Fine-tuned2025.12 | 61.6 | 65.6 | 57.6 | |
| LaserMode=Fine-tuned2025.12 | 61.6 | 65.6 | 56.8 | |
| R1-SearcherMode=Fine-tuned2025.12 | 56.8 | 56.8 | 48.8 | |
| Search-R1Mode=Fine-tuned2025.12 | 52.8 | 56.8 | 47.2 | |
| AutoRefineMode=Fine-tuned2025.12 | 48 | 53.6 | 42.4 |