List your company

End-to-end Task Completion on τ-Bench Retail, N=5

0.111Task Completion RateLlama3.1-8B
0.034040.054020.0740.09398Oct 8, 2025

Evaluation results

MethodLinks
Llama3.1-8BSchema configuration=PA-Tool2025.100.111
Qwen2.5-7BSchema configuration=PA-Tool2025.100.097
Llama3.1-8BSchema configuration=Base2025.100.097
Qwen2.5-7BSchema configuration=Base2025.100.068
Llama3.2-3BSchema configuration=PA-Tool2025.100.056
Llama3.2-3BSchema configuration=Base2025.100.045
Qwen2.5-3BSchema configuration=PA-Tool2025.100.038
Qwen2.5-3BSchema configuration=Base2025.100.037