Question Answering on LoCoMo Single-Hop
54.57F1 ScoreDeltaMem-8B-RL
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| DeltaMem-8B-RLTraining Protocol=RL-trained, Memory Manager Architecture=Agentic, Backbone=Qwen3-8B2026.04 | 54.57 | 43.4 | 82.05 | |
| DeltaMem-4B-RLTraining Protocol=RL-trained, Memory Manager Architecture=Agentic, Backbone=Qwen3-4B2026.04 | 53.96 | 43.05 | 80.02 | |
| DeltaMem-4o-miniTraining Protocol=Training-free, Memory Manager Architecture=Agentic, Backbone=GPT-4o-mini2026.04 | 53.75 | 42.51 | 80.38 | |
| DeltaMem-4BTraining Protocol=Training-free, Memory Manager Architecture=Agentic, Backbone=Qwen3-4B2026.04 | 49.79 | 40.87 | 75.62 | |
| ZepTraining Protocol=Training-free, Memory Manager Architecture=Multi-Agent2026.04 | 49.56 | 38.92 | 76.6 | |
| DeltaMem-8BTraining Protocol=Training-free, Memory Manager Architecture=Agentic, Backbone=Qwen3-8B2026.04 | 48.71 | 38.58 | 77.41 | |
| MemoryOSToken Usage=38742026.03 | 48.62 | 42.99 | — | |
| Mem0Token Usage=17642026.03 | 47.65 | 38.72 | — | |
| LightMemTraining Protocol=Training-free, Memory Manager Architecture=Multi-Agent2026.04 | 47.55 | 37.59 | 72.06 | |
| Memory-R1-7BTraining Protocol=RL-trained, Memory Manager Architecture=Multi-Agent, Backbone=7B2026.04 | 47.55 | 37.59 | 67.81 | |
| TA-MemToken Usage=37552026.03 | 44.87 | 38.74 | — | |
| A-MemToken Usage=25202026.03 | 44.65 | 37.06 | — | |
| Mem0Training Protocol=Training-free, Memory Manager Architecture=Multi-Agent2026.04 | 43.17 | 35.34 | 60.88 | |
| MemGPTToken Usage=169772026.03 | 41.04 | 34.34 | — | |
| LangMemTraining Protocol=Training-free, Memory Manager Architecture=Multi-Agent2026.04 | 40.91 | 33.63 | 71.12 | |
| LoCoMoToken Usage=169102026.03 | 40.36 | 29.05 | — | |
| A-MemTraining Protocol=Training-free, Memory Manager Architecture=Multi-Agent2026.04 | 33.34 | 27.58 | 54.05 | |
| ReadAgentToken Usage=6432026.03 | 9.67 | 7.66 | — | |
| MemoryBankToken Usage=4322026.03 | 6.61 | 5.16 | — |