Tool Calling on Stable Toolbench I1-Cat
76.07Pass RateParaTool
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| ParaToolBackbone=Qwen2.5-7B-Instruct2026.05 | 76.07 | 80.98 | |
| ToolLLaMABackbone=Qwen2.5-7B-Instruct2026.05 | 70.59 | 75.16 | |
| ParaToolBackbone=Llama3.1-8B-Instruct2026.05 | 69.94 | 73.01 | |
| Parameter+ReinvokeBackbone=Qwen2.5-7B-Instruct2026.05 | 68.71 | 74.23 | |
| ToolLLaMABackbone=Llama3.1-8B-Instruct2026.05 | 68.63 | 71.24 | |
| Context+Docs & ExamplesBackbone=Llama3.1-8B-Instruct2026.05 | 66.87 | 68.01 | |
| Parameter+ReinvokeBackbone=Llama3.1-8B-Instruct2026.05 | 65.64 | 68.01 | |
| Context+Docs & ExamplesBackbone=Qwen2.5-7B-Instruct2026.05 | 64.89 | 68.71 | |
| Parameter+ToolRetrieverBackbone=Llama3.1-8B-Instruct2026.05 | 63.19 | 66.26 | |
| Parameter+EmbsimBackbone=Llama3.1-8B-Instruct2026.05 | 60.74 | 63.9 | |
| Parameter+ToolRetrieverBackbone=Qwen2.5-7B-Instruct2026.05 | 59.9 | 63.8 | |
| Context+DocsBackbone=Llama3.1-8B-Instruct2026.05 | 59.51 | — | |
| Context+DocsBackbone=Qwen2.5-7B-Instruct2026.05 | 58.28 | — | |
| Parameter+EmbsimBackbone=Qwen2.5-7B-Instruct2026.05 | 57.67 | 62.58 | |
| Global ParameterizationBackbone=Qwen2.5-7B-Instruct2026.05 | 37.42 | 34.36 | |
| Global ParameterizationBackbone=Llama3.1-8B-Instruct2026.05 | 32.52 | 34.97 |