MemGPT addresses the challenge of managing context window limitations in LLM-powered applications. The RAG/retrieval-augmented generation approach allows for querying relevant context for LLM, and there are emerging infrastructure services related to context management and agents.
Table of contents
Managing LLM context like Operating System manages memoryehartford/dolphin-2.1-mistral-7bHuggingFaceH4/zephyr-7b-betaUsing local LLM OPEN_AI_KEY is not working · Issue #424 · cpacker/MemGPTImprove local LLM workflow (no more environment variables) · Issue #419 · cpacker/MemGPTMemGPTMemGPT: Overcoming Context Limitations for ChatGPT and Other LLMs for Document Chats & MoreUC Berkeley unveils MemGPT: Applying OS architecture to LLMs for unlimited contextMemGPT: Unlimited Memory without Token Constraints for Generative AI Platforms, like GPT-4, LaMDA…The difference of Assistant api and langchainPeering Inside GPT-4: Understanding Its Mixture of Experts (MoE) Architecture31 Impressions