Tool Calling on WHEN2TOOL Medium
-0.7Delta AccuracyNecessary (N)
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| Necessary (N)model averaging=six models2026.05 | -0.7 | -0.04 | -17.9 | |
| PROBE&PREFILLmodel averaging=six models, threshold (τ)=0.52026.05 | -3.4 | -0.54 | -6.2 | |
| Sparse (S)model averaging=six models2026.05 | -7.9 | -0.47 | -16.9 | |
| Necessary + Reason-then-Actmodel averaging=six models2026.05 | -16.3 | -0.82 | -19.7 | |
| Sparse + Reason-then-Actmodel averaging=six models2026.05 | -19.9 | -1.04 | -19.1 | |
| No Tool + Reason-then-Actmodel averaging=six models2026.05 | -23.5 | -1.12 | -20.9 | |
| No Tool (X)model averaging=six models2026.05 | -26.7 | -0.72 | -36.9 |