Memory on RULER SQuAD
81.25F1 Score (32K Context)MemAgent (Memory)
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| MemAgent (Memory)Model Category=Base and Task Models2026.01 | 81.25 | 77.34 | |
| RAM+Model Category=Merged Models2026.01 | 78.91 | 82.03 | |
| DARE+TAModel Category=Merged Models2026.01 | 77.34 | 70.31 | |
| DARE+TIESModel Category=Merged Models2026.01 | 76.56 | 74.22 | |
| TIESModel Category=Merged Models2026.01 | 75 | 75.78 | |
| RAMModel Category=Merged Models2026.01 | 74.22 | 79.69 | |
| TAModel Category=Merged Models2026.01 | 73.44 | 72.66 | |
| BaseModel Category=Base and Task Models2026.01 | 64.84 | 58.59 | |
| CURE (Coding)Model Category=Base and Task Models2026.01 | 60.94 | 44.22 | |
| ToolRL (Tool)Model Category=Base and Task Models2026.01 | 59.38 | 46.95 | |
| FisherModel Category=Merged Models2026.01 | 58.59 | 60.94 |