Tool Use on ToolBench Easy v1.0
97.9Tool Call Rate (TCR)NaviAgent
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| NaviAgentModel=Deepseek-V32025.06 | 97.9 | 71.8 | 4.4 | |
| NaviAgentModel=Qwen2.5-32B2025.06 | 88.1 | 61.1 | 4.29 | |
| α-UMIModel=Deepseek-V32025.06 | 80.8 | 59.7 | 5.95 | |
| α-UMIModel=Qwen2.5-32B2025.06 | 78.4 | 49.3 | 5.66 | |
| α-UMIModel=Qwen2.5-14B2025.06 | 77.7 | 39.2 | 5.53 | |
| NaviAgentModel=Qwen2.5-14B2025.06 | 64.2 | 50.3 | 4.18 | |
| ToolLLMModel=Deepseek-V32025.06 | 56.2 | 47.4 | 3.8 | |
| ToolLLMModel=Qwen2.5-14B2025.06 | 56.1 | 30.4 | 4.01 | |
| ReActModel=Deepseek-V32025.06 | 46.6 | 36.5 | 3.52 | |
| ToolLLMModel=Qwen2.5-32B2025.06 | 40.5 | 31.8 | 3.67 | |
| ReActModel=Qwen2.5-32B2025.06 | 33.1 | 25 | 3.5 | |
| ReActModel=Qwen2.5-14B2025.06 | 32.4 | 26.3 | 3.52 |