EvoMemState: State-Centric Long-Term Memory for Personal Agents
Zhishang Xiang · hf · 2026-11-06
EvoMemState, from Xiamen University with NTU, UBC, Waterloo and Jilin University, is a state-centric framework for personal agent long-term memory that fixes the blind spots of fact-centric paradigms. It has three components: ontology-guided memory state induction (persistent facts plus Relational States tracking validity, scope, and evolving relations), RL-based state evolution (internalizing memory-use experience into a request-specific Reasoning State), and state-aware reasoning. It consistently outperforms existing memory systems on retrieval completeness and memory-grounded response accuracy across long-term dialogue benchmarks. Code is open-sourced on GitHub.
More from coding & agent
- Building agents: life-oriented companions beat productivity tools in real use — yangyi · 2026-11-06
- ThinkyMachines' Inkling 2 Small trains models on its own: $426 of compute doubles Qwen score in 4.5 hours — Scobleizer · 2026-11-06
- Skillry ships a Claude Skill recreating Anthropic's homepage hero design — yihui_indie · 2026-11-06
- EdgeQuake: Rust GraphRAG with ingestion, knowledge graphs and RAG answers, 2.3k stars — JeremyCMorgan · 2026-11-06
- DataTalksClub launches free 5-week course on production AI agents for generative media — Al_Grigor · 2026-11-06
- Home motion capture in 15 minutes with GVHMR, Blender MCP and Claude — sidahuj · 2026-11-06