Tool-calling on Stable Toolbench I1-Inst
0.7909Pass RateParaTool
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| ParaToolBackbone=Qwen2.5-7B-Instruct2026.05 | 0.7909 | 0.8823 | |
| Context+Docs & ExamplesBackbone=Qwen2.5-7B-Instruct2026.05 | 0.7377 | 0.7712 | |
| ToolLLaMABackbone=Qwen2.5-7B-Instruct2026.05 | 0.724 | 0.681 | |
| Parameter+ReinvokeBackbone=Qwen2.5-7B-Instruct2026.05 | 0.7059 | 0.7581 | |
| Context+DocsBackbone=Qwen2.5-7B-Instruct2026.05 | 0.7049 | — | |
| Parameter+ToolRetrieverBackbone=Qwen2.5-7B-Instruct2026.05 | 0.6928 | 0.7362 | |
| ParaToolBackbone=Llama3.1-8B-Instruct2026.05 | 0.6797 | 0.6993 | |
| Parameter+EmbsimBackbone=Qwen2.5-7B-Instruct2026.05 | 0.6536 | 0.7189 | |
| ToolLLaMABackbone=Llama3.1-8B-Instruct2026.05 | 0.6442 | 0.6258 | |
| Context+Docs & ExamplesBackbone=Llama3.1-8B-Instruct2026.05 | 0.6275 | 0.6536 | |
| Parameter+ReinvokeBackbone=Llama3.1-8B-Instruct2026.05 | 0.6209 | 0.634 | |
| Parameter+ToolRetrieverBackbone=Llama3.1-8B-Instruct2026.05 | 0.6013 | 0.5668 | |
| Parameter+EmbsimBackbone=Llama3.1-8B-Instruct2026.05 | 0.5425 | 0.5686 | |
| Context+DocsBackbone=Llama3.1-8B-Instruct2026.05 | 0.5163 | — | |
| Global ParameterizationBackbone=Qwen2.5-7B-Instruct2026.05 | 0.4248 | 0.3987 | |
| Global ParameterizationBackbone=Llama3.1-8B-Instruct2026.05 | 0.3464 | 0.3791 |