List your company

Tool-calling on Stable Toolbench I3-Inst

68.85Pass RateParaTool
27.66638.35849.0559.742May 28, 2026

Evaluation results

MethodLinks
ParaToolBackbone=Qwen2.5-7B-Instruct2026.0568.8573.77
ParaToolBackbone=Llama3.1-8B-Instruct2026.0565.5768.85
Context+Docs & ExamplesBackbone=Qwen2.5-7B-Instruct2026.0565.5768.85
ToolLLaMABackbone=Qwen2.5-7B-Instruct2026.0565.5770.49
Context+Docs & ExamplesBackbone=Llama3.1-8B-Instruct2026.0563.9355.74
ToolLLaMABackbone=Llama3.1-8B-Instruct2026.0560.6657.38
Parameter+ReinvokeBackbone=Qwen2.5-7B-Instruct2026.0560.6662.3
Context+DocsBackbone=Qwen2.5-7B-Instruct2026.0558.82—
Parameter+ReinvokeBackbone=Llama3.1-8B-Instruct2026.0557.3855.74
Context+DocsBackbone=Llama3.1-8B-Instruct2026.0552.46—
Parameter+ToolRetrieverBackbone=Qwen2.5-7B-Instruct2026.0552.4655.74
Parameter+ToolRetrieverBackbone=Llama3.1-8B-Instruct2026.0550.8252.46
Parameter+EmbsimBackbone=Qwen2.5-7B-Instruct2026.0550.8254.1
Parameter+EmbsimBackbone=Llama3.1-8B-Instruct2026.0547.5444.26
Global ParameterizationBackbone=Qwen2.5-7B-Instruct2026.0531.3327.87
Global ParameterizationBackbone=Llama3.1-8B-Instruct2026.0529.2521.31