Tagged articles
LLM memory
14 articles · Page 1 of 1
EM²Mem introduces an event-centric multimodal memory system that organizes experiences by events, aligns multimodal cues during memory construction, and enables efficient retrieval, outperforming baselines on long-video QA benchmarks while reducing inference time and token usage.
EMNLP 2026EM²MemLLM memory
0 likes · 13 min read

The article evaluates MemT-1.0, the first open-source memory-native LLM, testing its 36B Nano version on an A100 GPU by injecting four strict financial payment rules into its internal memory, then verifying via AST analysis that the model recalls and applies those rules in a new session with zero context, while detailing architecture, benchmarks, and limitations.
AST verificationGDN-2LLM memory
0 likes · 25 min read

VoiceMem introduces a brain-inspired dual-memory system for real-time voice AI, separating factual and emotional memory to achieve 134ms retrieval latency and enable long-term personalized companionship.
Emotional AILLM memoryMemory Systems
0 likes · 11 min read


ArchitectAug 9, 2026 · Artificial Intelligence
This article reviews the recent “Memory for Large Language Models” survey, outlining three orthogonal design axes—representation, update dynamics, and persistence—and maps them to engineering concerns such as work‑set, compressed state, long‑term items, and raw evidence, while discussing evaluation dimensions and practical implementation guidelines for agent systems.
Agent ArchitectureAttentionLLM memory
0 likes · 20 min read

SkillOpt‑Sleep, the deployment companion for Microsoft’s open‑source SkillOpt, reviews offline Claude or Codex sessions, extracts coding habits and project conventions, validates them through a gate, and writes learned rules into protected blocks, enabling the assistant to remember and improve over time.
AI coding assistantClaude CodeCodex
0 likes · 26 min read

δ‑mem introduces a compact 8×8 online state matrix that, without expanding context windows or altering the Transformer backbone, provides effective long‑term memory for large language models, achieving up to 1.31× performance gains on memory‑intensive tasks while adding only 0.12% parameters.
LLM memoryTransformerdelta-mem
0 likes · 15 min read

This article explains why LLMs are stateless, distinguishes short‑term from long‑term memory needs for agents, critiques common memory solutions, and then details Claude Code’s two‑layer architecture—static CLAUDE.md with six hierarchical files and a dynamic auto‑memory system that uses structured markdown, a lightweight selector model, and aging warnings—to provide a practical, source‑level blueprint for building robust agent memory.
Agent ArchitectureClaude CodeDynamic Memory
0 likes · 33 min read

Claude Code achieves a 39% performance gain in long‑term memory for LLM agents by replacing complex vector stores and graph databases with a simple folder of markdown files, leveraging context editing and tool calls that reduce token usage by 84% while keeping memory transparent and version‑controlled.
AnthropicClaude CodeLLM memory
0 likes · 11 min read

The paper introduces the Dynamic Memory Forest (DMF) framework, inspired by human memory consolidation and growth, which transforms fragmented long‑term dialogue histories into structured memory trees, enabling entropy‑driven walks and grafting mechanisms that markedly improve coherence and efficiency of LLM responses.
Dynamic Memory ForestEntropy-Driven WalkLLM memory
0 likes · 11 min read

This article details the technical evaluation and deep integration of LanceDB as a memory plugin for the OpenClaw‑based ArkClaw agent platform, covering plugin selection, core enhancements such as mixed retrieval, hierarchical memory, Autodream processing, Context Engine optimizations, Git‑style version control, and the vision of a unified edge‑cloud memory lake.
AI agentsArkClawLLM memory
0 likes · 12 min read

The article analyzes how extending the context window of large language models leads to rapid forgetting, outlines seven concrete failure modes, examines cognitive‑science‑based memory architectures, and walks through practical layers—from Python lists to markdown files to vector retrieval—highlighting why simple context expansion alone cannot solve the problem.
Context WindowLLM memoryagent design
0 likes · 10 min read

The article explains how Agent Harnesses tightly bind an LLM agent's memory, why using closed‑source Harnesses hands control of that memory to third‑party APIs, and argues that open‑source Harnesses are essential for retaining data ownership and avoiding lock‑in.
AI agentsAgent HarnessLLM memory
0 likes · 12 min read

The article analyzes Anthropic's guidance on building effective AI agents, contrasts workflow and agent architectures, outlines criteria for choosing agents, presents six incremental design patterns, and shares practical principles such as simplicity, transparency, and robust tool interfaces.
AI agentsLLM memoryagent design
0 likes · 9 min read

The article explains how Baidu's commercial advertising platform leverages generative AI agents—covering their core capabilities of understanding, planning, execution, and persona—to overcome challenges such as hallucination and integration, describing a multi‑layer architecture, key technologies, real‑world case studies, and the resulting performance and operational benefits.
AI agentsAdvertising PlatformLLM memory
0 likes · 18 min read
