Tool selection on TaskBench
21.1F1 ScoreCLIN
Evaluation Results
| Method | Links | |
|---|---|---|
| CLINBackbone=Qwen3-4B2026.05 | 21.1 | |
| OLIVIABackbone=Mistral-7B-v0.12026.05 | 20.8 | |
| OLIVIABackbone=Qwen3-4B2026.05 | 20.5 | |
| CLINBackbone=Mistral-7B-v0.12026.05 | 19.3 | |
| ReActBackbone=Mistral-7B-v0.12026.05 | 17.8 | |
| ReActBackbone=Qwen3-4B2026.05 | 17.4 | |
| BanditBackbone=Qwen3-4B2026.05 | 15.5 | |
| BanditBackbone=Mistral-7B-v0.12026.05 | 15.5 | |
| BM25Backbone=Qwen3-4B2026.05 | 11.6 | |
| BM25Backbone=Mistral-7B-v0.12026.05 | 11.6 | |
| CoTBackbone=Mistral-7B-v0.12026.05 | 3.6 | |
| CoTBackbone=Qwen3-4B2026.05 | 3 |