A Heterogeneous Temporal Memory Governance Framework for Long-Term LLM Persona Consistency

Large language models often suffer from fact loss, timeline confusion, persona drift, and reduced stability during long-range interaction, especially under high-noise knowledge bases, context clearing, and cross-model transfer. To address these issues, we introduce ARPM, an external temporal memory ...