Language Agent Memory Management on PersonaMem
88.64Recall FactsRGMem
Evaluation Results
| Method | Links | ||||||||
|---|---|---|---|---|---|---|---|---|---|
| RGMemBackbone=GPT-4.12025.10 | 88.64 | 35.04 | 87.03 | 70.89 | 83.02 | 72.55 | 85.07 | 74.01 | |
| A-MemBackbone=GPT-4.12025.10 | 82.57 | 28.31 | 86.35 | 56.72 | 70.12 | 65.13 | 77.81 | 63.95 | |
| Mem0Backbone=GPT-4.12025.10 | 81.02 | 16.28 | 81.82 | 54.33 | 54.35 | 65.13 | 81.22 | 60.44 | |
| Memory OSBackbone=GPT-4.12025.10 | 79.72 | 19.33 | 82.16 | 60.34 | 73.66 | 74.18 | 77.81 | 65.03 | |
| LangMemBackbone=GPT-4.12025.10 | 77.82 | 24.27 | 82.16 | 54.33 | 42.33 | 63.22 | 70.59 | 58.23 | |
| RGMemBackbone=GPT-4o-mini2025.10 | 77.06 | 26.47 | 85.29 | 67.82 | 73.66 | 56.62 | 75.47 | 63.87 | |
| Memory OSBackbone=GPT-4o-mini2025.10 | 72.59 | 22.62 | 84.42 | 69.11 | 67.82 | 35.72 | 64.71 | 54.23 | |
| Mem0Backbone=GPT-4o-mini2025.10 | 69.57 | 19.53 | 78.75 | 56.72 | 61.82 | 52.19 | 68.25 | 56.79 | |
| LangMemBackbone=GPT-4o-mini2025.10 | 64.76 | 23.93 | 81.82 | 53.3 | 45.45 | 42.26 | 58.12 | 52.36 | |
| LLM (Vanilla)Backbone=GPT-4.12025.10 | 64.76 | 19.53 | 81.82 | 67.21 | 53.14 | 53.91 | 51.62 | 51.86 | |
| A-MemBackbone=GPT-4o-mini2025.10 | 59.78 | 9.22 | 80.19 | 53.3 | 44.32 | 45.42 | 53.25 | 49.17 | |
| LLM (Vanilla)Backbone=GPT-4o-mini2025.10 | 55.34 | 12.33 | 70.02 | 58.21 | 42.33 | 33.05 | 34.31 | 39.21 |