Tool selection on BFCL
36.6F1 ScoreOLIVIA
Evaluation Results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| OLIVIABackbone=Qwen3-4B2026.05 | 36.6 | — | — | — | — | |
| OLIVIABackbone=Mistral-7B-v0.12026.05 | 35.2 | — | — | — | — | |
| CLINBackbone=Mistral-7B-v0.12026.05 | 32.4 | — | — | — | — | |
| CLINBackbone=Qwen3-4B2026.05 | 30.7 | — | — | — | — | |
| ReActBackbone=Mistral-7B-v0.12026.05 | 28.1 | — | — | — | — | |
| ReActBackbone=Qwen3-4B2026.05 | 27.2 | — | — | — | — | |
| BanditBackbone=Qwen3-4B2026.05 | 24.2 | — | — | — | — | |
| BanditBackbone=Mistral-7B-v0.12026.05 | 24.2 | — | — | — | — | |
| BM25Backbone=Qwen3-4B2026.05 | 18.1 | — | — | — | — | |
| BM25Backbone=Mistral-7B-v0.12026.05 | 18.1 | — | — | — | — | |
| CoTBackbone=Mistral-7B-v0.12026.05 | 10.7 | — | — | — | — | |
| CoTBackbone=Qwen3-4B2026.05 | 8.2 | — | — | — | — | |
| F1Scorer=BM25, N=3702026.05 | 0.889 | — | — | — | 6.4 | |
| F1Scorer=embed, N=3702026.05 | 0.858 | — | — | — | 1.6 | |
| Best Fixed-KScorer=BM25, N=3702026.05 | — | — | 0.908 | — | — | |
| Best Fixed-KScorer=embed, N=3702026.05 | — | — | 0.975 | — | — | |
| BoRScorer=BM25, N=3702026.05 | — | 0.903 | — | 7.4 | — | |
| BoRScorer=embed, N=3702026.05 | — | 0.85 | — | 1.4 | — |