Information Retrieval on STaRK PRIME human-generated (test)
57.1Hit@1AF-Retriever
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| AF-Retriever2025.05 | 57.1 | 69.4 | 62.7 | |
| 4StepFocus2025.05 | 50.5 | 65.5 | 57.9 | |
| KAR2025.05 | 45 | 60.6 | 51.9 | |
| AvaTaR2025.05 | 33 | 51.4 | 41 | |
| VSS + Reranker2025.05 | 28.4 | 48.6 | 36.2 | |
| VSSembedding_model=Multi-ada-0022025.05 | 24.5 | 39.8 | 33 | |
| ReAct2025.05 | 21.7 | 33.3 | 28.2 | |
| VSSembedding_model=Ada-0022025.05 | 17.4 | 34.7 | 26.4 | |
| Reflexion2025.05 | 16.5 | 33 | 24 | |
| DPR2025.05 | 2 | 9.2 | 7 |