Tool-use Search Evaluation on SLATE synthetic
68.5Tool Match RateEGB-Sampling (Ours)
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| EGB-Sampling (Ours)Model=Claude-Sonnet-42026.04 | 68.5 | 54 | 87.9 | |
| ReActModel=Claude-Sonnet-42026.04 | 66.4 | 29.3 | 85.7 | |
| Baseline-LLMModel=Claude-Sonnet-42026.04 | 65.2 | — | — | |
| LATSModel=Claude-Sonnet-42026.04 | 63.4 | 36.5 | 87.3 | |
| ReflexionModel=Claude-Sonnet-42026.04 | 61.5 | 44.7 | 83.4 |