Knowledge Pair-wise Diversity on DailyDialog (test)
89.58PrecisionInstructor
Evaluation Results
| Method | Links | ||||||
|---|---|---|---|---|---|---|---|
| Instructor2025.11 | 89.58 | 89.76 | 89.62 | 27.97 | 13.88 | 26.76 | |
| GTE2025.11 | 89.57 | 89.54 | 89.53 | 28.49 | 11.33 | 26.73 | |
| SBERT2025.11 | 88.85 | 88.84 | 88.83 | 30.43 | 12.99 | 28.06 | |
| Contriever2025.11 | 88.62 | 88.45 | 88.51 | 25.28 | 9.33 | 22.95 | |
| ChatGPT Inference2025.11 | 88.38 | 87.85 | 88.11 | 22.82 | 6.03 | 19.35 | |
| DPR2025.11 | 86.75 | 86.79 | 86.75 | 18.05 | 4.7 | 17.31 | |
| MCTS-Driven Knowledge RetrievalConcept Bridging=false2025.11 | 86.37 | 86.39 | 86.38 | 21.94 | 6.65 | 17.84 | |
| MCTS-Driven Knowledge RetrievalReasoner=false2025.11 | 83.77 | 83.71 | 83.72 | 13.22 | 2.22 | 11.72 | |
| MCTS-Driven Knowledge Retrieval2025.11 | 83.63 | 83.68 | 83.64 | 10.29 | 1.14 | 9.16 |