Skill retrieval on ToolQA
56.4nDCG@1Qwen3-235B
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| Qwen3-235BRetrieval Stage=Reranker, Candidate selection=top-50 candidates retrieved by BM252026.04 | 56.4 | 68.1 | |
| Mistral3.1-24BRetrieval Stage=Reranker, Candidate selection=top-50 candidates retrieved by BM252026.04 | 53.9 | 67.3 | |
| Llama-3.3-70BRetrieval Stage=Reranker, Candidate selection=top-50 candidates retrieved by BM252026.04 | 51 | 66.2 | |
| Qwen3-32BRetrieval Stage=Reranker, Candidate selection=top-50 candidates retrieved by BM252026.04 | 43.7 | 62.7 | |
| Qwen3-4BRetrieval Stage=Reranker, Candidate selection=top-50 candidates retrieved by BM252026.04 | 39.9 | 57.5 | |
| BGERetrieval Stage=First-stage2026.04 | 32.2 | 58.6 | |
| Llama-3.1-8BRetrieval Stage=Reranker, Candidate selection=top-50 candidates retrieved by BM252026.04 | 25.7 | 47.5 | |
| ContrieverRetrieval Stage=First-stage2026.04 | 21.2 | 31.9 | |
| BM25Retrieval Stage=First-stage2026.04 | 7 | 27 | |
| TF-IDFRetrieval Stage=First-stage2026.04 | 7 | 18.9 | |
| HybridRetrieval Stage=First-stage2026.04 | 7 | 44.5 |