Function Calling on Tool-call benchmark (n=200)
100Accuracy (12-char)TA-Fast-Semantic
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| TA-Fast-SemanticBudget (K)=32, Sample size (n)=200, Deployment scenario=Function-calling2026.04 | 100 | 100 | 100 | |
| TA-Fast-RegexBudget (K)=32, Sample size (n)=200, Deployment scenario=Function-calling2026.04 | 94 | 88 | 91 | |
| TOVABudget (K)=32, Sample size (n)=200, Deployment scenario=Function-calling2026.04 | 0 | 0 | 0 |