Tool Use on ToolBench Medium v1.0
96.3TCRNaviAgent
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| NaviAgentModel=Deepseek-V32025.06 | 96.3 | 48.5 | 4.45 | |
| α-UMIModel=Deepseek-V32025.06 | 89.4 | 32.9 | 5.95 | |
| NaviAgentModel=Qwen2.5-32B2025.06 | 81.7 | 41.5 | 4.6 | |
| α-UMIModel=Qwen2.5-32B2025.06 | 78.8 | 26 | 6.02 | |
| α-UMIModel=Qwen2.5-14B2025.06 | 77.9 | 25 | 5.88 | |
| NaviAgentModel=Qwen2.5-14B2025.06 | 60.1 | 32.3 | 4.38 | |
| ToolLLMModel=Deepseek-V32025.06 | 58.8 | 30 | 3.92 | |
| ReActModel=Deepseek-V32025.06 | 58.7 | 38.5 | 3.5 | |
| ToolLLMModel=Qwen2.5-14B2025.06 | 53.8 | 19.7 | 4.02 | |
| ToolLLMModel=Qwen2.5-32B2025.06 | 48.6 | 30.3 | 3.85 | |
| ReActModel=Qwen2.5-32B2025.06 | 35.6 | 24.5 | 3.6 | |
| ReActModel=Qwen2.5-14B2025.06 | 24.5 | 16.8 | 3.67 |