Property-based retrieval on Responses (test)
48.5MAPCombined
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| CombinedLLM reranking=true, n_phrases=18, T=0.22025.12 | 48.5 | 93.4 | |
| CombinedLLM reranking=false, n_phrases=18, T=0.22025.12 | 47.6 | 84.2 | |
| SAELLM reranking=true, n_phrases=18, T=0.22025.12 | 42.8 | 88.4 | |
| SAELLM reranking=false, n_phrases=18, T=0.22025.12 | 41.8 | 76.4 | |
| OpenAI+LLMLLM reranking=true, n_phrases=18, T=0.22025.12 | 38.7 | 90.4 | |
| OpenAI+LLMLLM reranking=false, n_phrases=18, T=0.22025.12 | 37.3 | 72.2 |