Tool Sequence Recommendation on ToolBench (test)
27.1Set-F1Hybrid Sem-Graph
Evaluation Results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| Hybrid Sem-GraphMethod Type=Single-Stage2026.04 | 27.1 | 13 | 0.044 | 8.2 | 26.2 | |
| GS-Hybrid + Sem-SortMethod Type=Two-Stage, Stage 1=GS-Hybrid2026.04 | 27.1 | 13.3 | 0.051 | 8.3 | 26.2 | |
| GS-Hybrid + Hyb-RerankMethod Type=Two-Stage, Stage 1=GS-Hybrid2026.04 | 27.1 | 15.2 | 0.089 | 9.7 | 25.2 | |
| GS-Hybrid + Opt-PermMethod Type=Two-Stage, Stage 1=GS-Hybrid2026.04 | 27.1 | 15.1 | 0.086 | 9.7 | 22.7 | |
| GS-Hybrid + LR (ours)Method Type=Two-Stage, Stage 1=GS-Hybrid2026.04 | 27.1 | 15.6 | 0.096 | 9.8 | 23.4 | |
| Semantic OnlyMethod Type=Single-Stage2026.04 | 26.6 | 12.3 | 0.042 | 7 | 26.2 | |
| Beam SearchMethod Type=Single-Stage2026.04 | 24 | 14.9 | 0.089 | 9.9 | 20.6 | |
| BM25Method Type=Single-Stage2026.04 | 21.5 | 7.3 | 0.001 | 4.1 | 14.3 |