Research map / Architectures and efficiency / RNNs and normalization
Neuromem: A Granular Decomposition of the Streaming Lifecycle in External Memory for LLMs
Ruicheng Zhang, Xinyi Li, Tianyi Xu, Shuhao Zhang, Xiaofei Liao, Hai Jin
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add Rewrite (triplet extraction) normalization (ablation) | Queue+Segment · LoCoMo R5 | token-level F1: 0.325 -> 0.112 | Hurt |
| add Rewrite normalization (ablation) | Property Graph · LoCoMo R5 | token-level F1: 0.301 -> 0.113 | Hurt |
| add Rewrite normalization with Qwen-Plus and 10 triplets (ablation) | Property Graph · LoCoMo subset | mean F1: 0.389 -> 0.163 | Hurt |
| add Enrich (summarization) normalization (ablation) | Queue+Segment · LoCoMo R5 | token-level F1: 0.325 -> 0.324 | No clear effect |
| add CRUD LLM conflict resolution (ablation) | Property Graph · LoCoMo | mean F1: 0.300 -> 0.301 | No clear effect |
| add Decay Eviction consolidation (ablation) | Queue+Segment · LoCoMo | mean F1: 0.341 -> 0.342 | No clear effect |
| add Forgetting consolidation (ablation) | LSH Hash · LoCoMo | mean F1: 0.101 -> 0.101 | No clear effect |
| add Keyword query extraction (ablation) | Queue+Segment · LoCoMo | mean F1: 0.341 -> 0.099 | Hurt |
| add Decompose query formulation (ablation) | Queue+Segment · LoCoMo | mean F1: 0.341 -> 0.328 | Hurt |
| add Decompose query formulation (ablation) | LSH Hash · LoCoMo | mean F1: 0.063 -> 0.094 | Improved |
| add Multi-query context integration (ablation) | Property Graph · LoCoMo | mean F1: 0.299 -> 0.299 | No clear effect |
| add Multi-query context integration (ablation) | LSH Hash · LoCoMo | mean F1: 0.097 -> 0.092 | Hurt |
Closest papers by meaning
- MEMAUDIT: An Exact Package-Oracle Evaluation Protocol for Budgeted Long-Term LLM Memory Writing (NeurIPS 2026 · similarity 0.88)
- SimpleMem: Efficient Lifelong Memory for LLM Agents (ICML 2026 · similarity 0.87)
- Peripheral Memory for LLMs: Integration of Sequential Memory Banks with Adaptive Querying (ICML 2025 · similarity 0.87)
- Frozen Memory Is Not Enough: Rethinking External Memory as Extraction (NeurIPS 2026 · similarity 0.87)
- UMEM: Unified Memory Extraction and Management Framework for Generalizable Memory (ICML 2026 · similarity 0.86)
- Quantifying Memory Utilization with Effective State-Size (ICML 2025 · similarity 0.86)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
