LifeFuse-Mem: Lifecycle-Aware State Fusion Against Temporary Overwriting for Long-Term Memory
Zhao and colleagues introduce a memory architecture that labels each write with its intended lifetime, so information meant to apply only to the current context cannot overwrite knowledge meant to persist.
Ask this paper
The failure mode is specific. A long-running agent that stores transient context the same way it stores durable facts will let temporary information overwrite stable knowledge, which shows up later as behavioral drift.
Lifecycle metadata at write time, phase-aware readout at evaluation. Write episodes carry lifecycle labels during training, and reads are conditioned on phase, which lets the system separate what should remain influential from what should not.
Separate components per temporal commitment. LifeFuse-Mem gives stable and transient knowledge dedicated memory components with lifecycle-aware updates, so each evolves locally without transient context being promoted to durable state.
Results on a controlled benchmark. On an anti-overwrite benchmark built for this setting, the method improves acquisition-controlled retention and reduces temporary overwrite.
The authors state the limit. On two public long-memory benchmarks the method is competitive rather than dominant, so the contribution is the diagnosis of overwrite and an explicit signal for mitigating it.
Abstract
Long-running LLM agents require memory mechanisms that maintain coherent internal states across interactions. We study a lifecycle-labeled memory setting in which write episodes provide lifecycle metadata during training, and phase-aware readout is used during evaluation. This setting reflects the need to distinguish information that should remain influential across future interactions from information that should affect only the current context. A mismatch between these lifecycles can cause temporary information to overwrite durable knowledge, leading to behavioral drift in persistent agents. Within this setting, we introduce \textbf{LifeFuse-Mem}, a lifecycle-aware neural memory framework that separates information according to its temporal commitment. LifeFuse-Mem uses dedicated memory components and lifecycle-aware updates to allow stable and transient knowledge to evolve locally without converting temporary context into durable state. On the controlled anti-overwrite benchmark, LifeFuse-Mem improves acquisition-controlled retention and reduces temporary overwrite; on two public long-memory benchmarks, it remains broadly competitive. These results suggest that explicit lifecycle signals can help diagnose and mitigate overwrite in compact online memory.