Cross-application Task Execution on LibreOffice Calc 45 tasks
46.7Pass RateExperience Bank
Evaluation results
| Method | Links | |
|---|---|---|
| Experience BankEvolution Strategy=Experience Augmentation2026.04 | 46.7 | |
| BaselineEvolution Strategy=None2026.04 | 44.4 | |
| Distillation + Experience BankEvolution Strategy=Trajectory Distillation + Experience Augmentation2026.04 | 42.2 |