Agent Task and Scenario Completion on AppWorld (dev)
89.5Task Goal CompletionTrajectory-Informed Memory Generation
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| Trajectory-Informed Memory GenerationMemory Configuration=+Memory2026.03 | 89.5 | 73.7 | |
| BaselineMemory Configuration=Baseline2026.03 | 77.2 | 47.4 |