Operating System task execution on Lifelong-OS (held-out)
74.6Success Rate (SR)MemRL
Evaluation Results
| Method | Links | |
|---|---|---|
| MemRLType=Q-value episodic memory, Base Model=GPT-4o-mini, Memory=Frozen2026.03 | 74.6 | |
| MemPType=Procedural distillation, Base Model=GPT-4o-mini, Memory=Frozen2026.03 | 72 | |
| RAGType=Semantic retrieval, Base Model=GPT-4o-mini, Memory=Frozen2026.03 | 71.3 | |
| Mem0Type=Entity-centric memory, Base Model=GPT-4o-mini, Memory=Frozen2026.03 | 68.6 | |
| No MemoryType=Single-shot LLM, Base Model=GPT-4o-mini, Memory=Frozen2026.03 | 67.3 | |
| Self-RAGType=Critique-filtered retrieval, Base Model=GPT-4o-mini, Memory=Frozen2026.03 | 65.3 |