{"@context":"https://schema.org","@type":"CreativeWork","@id":"https://forgecascade.org/public/capsules/0548e603-3763-40b7-87ea-739d92d68f82","name":"LLM Memory Systems: Context Windows, External Memory, and Retrieval","text":"Context window limits: GPT-4 128k, Claude 3.5 200k, Gemini 1.5 1M tokens. Long-context challenges: lost in the middle (Liu 2023), attention dilution, quadratic compute cost. External memory architectures: RAG (non-parametric), MemGPT (paging), Zep (temporal graph memory), LangMem. Episodic memory: store past interactions as searchable events. Semantic memory: factual knowledge base (knowledge graph). Working memory: in-context window. Procedural memory: embedded in model weights. Forgetting curves: Ebbinghaus for humans, confidence decay for AI capsules. Forge implements confidence decay (0.999/day) to flag stale knowledge for re-verification — directly addressing AI memory degradation.","keywords":["llm","memory","context-window"],"about":[],"citation":[],"isPartOf":{"@type":"Dataset","name":"Forge Cascade Knowledge Graph","url":"https://forgecascade.org"},"publisher":{"@type":"Organization","name":"Forge Cascade","url":"https://forgecascade.org"}}