Tool Selection on ToolBench
58.5F1 ScoreOLIVIA
Evaluation Results
| Method | Links | |
|---|---|---|
| OLIVIABackbone=Qwen3-4B2026.05 | 58.5 | |
| CLINBackbone=Qwen3-4B2026.05 | 55.3 | |
| ReActBackbone=Qwen3-4B2026.05 | 52 | |
| OLIVIABackbone=Mistral-7B-v0.12026.05 | 42 | |
| CLINBackbone=Mistral-7B-v0.12026.05 | 39.4 | |
| ReActBackbone=Mistral-7B-v0.12026.05 | 38.6 | |
| BM25Backbone=Qwen3-4B2026.05 | 35.3 | |
| BM25Backbone=Mistral-7B-v0.12026.05 | 35.3 | |
| BanditBackbone=Qwen3-4B2026.05 | 34.8 | |
| BanditBackbone=Mistral-7B-v0.12026.05 | 34.8 | |
| CoTBackbone=Qwen3-4B2026.05 | 21.1 | |
| CoTBackbone=Mistral-7B-v0.12026.05 | 14.2 |