List your company

Tool-calling on Stable Toolbench I1-Inst

0.7909Pass RateParaTool
0.328620.4486350.568650.688665May 28, 2026

Evaluation results

MethodLinks
ParaToolBackbone=Qwen2.5-7B-Instruct2026.050.79090.8823
Context+Docs & ExamplesBackbone=Qwen2.5-7B-Instruct2026.050.73770.7712
ToolLLaMABackbone=Qwen2.5-7B-Instruct2026.050.7240.681
Parameter+ReinvokeBackbone=Qwen2.5-7B-Instruct2026.050.70590.7581
Context+DocsBackbone=Qwen2.5-7B-Instruct2026.050.7049—
Parameter+ToolRetrieverBackbone=Qwen2.5-7B-Instruct2026.050.69280.7362
ParaToolBackbone=Llama3.1-8B-Instruct2026.050.67970.6993
Parameter+EmbsimBackbone=Qwen2.5-7B-Instruct2026.050.65360.7189
ToolLLaMABackbone=Llama3.1-8B-Instruct2026.050.64420.6258
Context+Docs & ExamplesBackbone=Llama3.1-8B-Instruct2026.050.62750.6536
Parameter+ReinvokeBackbone=Llama3.1-8B-Instruct2026.050.62090.634
Parameter+ToolRetrieverBackbone=Llama3.1-8B-Instruct2026.050.60130.5668
Parameter+EmbsimBackbone=Llama3.1-8B-Instruct2026.050.54250.5686
Context+DocsBackbone=Llama3.1-8B-Instruct2026.050.5163—
Global ParameterizationBackbone=Qwen2.5-7B-Instruct2026.050.42480.3987
Global ParameterizationBackbone=Llama3.1-8B-Instruct2026.050.34640.3791