Tool selection on API-Bench
96.3AccuracyMTA
Evaluation Results
| Method | Links | |
|---|---|---|
| MTABackbone=Qwen1.5-Chat (7B)2026.01 | 96.3 | |
| FTBackbone=Qwen1.5-Chat (7B)2026.01 | 95.11 | |
| FTBackbone=Baichuan2-Chat (7B)2026.01 | 94.77 | |
| MTABackbone=Chinese Alpaca-2 (7B)2026.01 | 93.8 | |
| MTABackbone=Baichuan2-Chat (7B)2026.01 | 93.3 | |
| FTBackbone=Chinese Alpaca-2 (7B)2026.01 | 88.34 | |
| FTBackbone=ChatGLM3 (6B)2026.01 | 83.12 | |
| MTABackbone=ChatGLM3 (6B)2026.01 | 82.45 | |
| ReActBackbone=Qwen1.5-Chat (7B)2026.01 | 70.33 | |
| ReActBackbone=ChatGLM3 (6B)2026.01 | 65.34 | |
| CoTBackbone=Qwen1.5-Chat (7B)2026.01 | 59.1 | |
| CoTBackbone=ChatGLM3 (6B)2026.01 | 55.12 | |
| ReActBackbone=Baichuan2-Chat (7B)2026.01 | 51.79 | |
| BaseBackbone=ChatGLM3 (6B)2026.01 | 50.23 | |
| CoTBackbone=Baichuan2-Chat (7B)2026.01 | 46.34 | |
| BaseBackbone=Qwen1.5-Chat (7B)2026.01 | 45.02 | |
| BaseBackbone=Baichuan2-Chat (7B)2026.01 | 42.48 | |
| ReActBackbone=Chinese Alpaca-2 (7B)2026.01 | 39.12 | |
| CoTBackbone=Chinese Alpaca-2 (7B)2026.01 | 38.91 | |
| BaseBackbone=Chinese Alpaca-2 (7B)2026.01 | 34.67 |