AI Wiki — Knowledge Base

The Log is the Agent 它的核心观点是：不要把日志当成 agent 的副产品，而要把 append-only event log 当成 agent 本身。在 ActiveGraph 里，目标、规则、工具调用、LLM 响应、产物和关系都写入同一条事件日志；当前状…

The Log is the Agent 它的核心观点是：不要把日志当成 agent 的副产品，而要把 append-only event log 当成 agent 本身。在 ActiveGraph 里，目标、规则、工具调用、LLM 响应、产物和关系都写入同一条事件日志；当前状态只是日志 replay 后投影出来的一张图。这样带来三个关键能力：可确定性重放可在任意事件处分叉实验从最终输出

agentsyohei-nakajima--x

5 Jul 2026

Search: “llm”

FreqDepthKV: Frequency-Guided Depth Sharing for Robust KV Cache Compression in Long-Context LLM Inference

From Blueprint to Reality: Modeling and Applying Putnam's Social Capital Theory with LLM-based Multi-agent Simulations

Highly-recommended read. Aligns with what I see in my own harness: > Pi harness got the same success rate as harnesses from the LLM vendors …

IndoorR2X: Indoor Robot-to-Everything Coordination with LLM-Driven Planning

Information Gain-based Rollout Policy Optimization: An Adaptive Tree-Structured Rollout Approach for Multi-Turn LLM Agents

LLM Agents for Deliberative Collaboration: A Study on Joint Decision Making Under Partial Observability

MASCA: LLM based-Multi Agents System for Credit Assessment

More Convincing, Not More Correct: Self-Play Reward Hacking of Reference-Free LLM Judges

Most LLM Conformity Needs No Speaker: Measuring the Speaker-Free Floor in Peer-Pressure Benchmarks

Think Before You Grid-Search: Floor-First Triage for LLM Serving

We're releasing ZML/LLMD, our homegrown LLM server built on top of our homegrown high performance heterogeneous inference stack. It ships wi…

Detecting Answer-Driven Reasoning in LLM-Based Educational Tutors via Truncated Chain-of-Thought Auditing

Elastic Gang: Per-Token Membership Change for a Hard-Barriered LLM Inference Gang Co-Scheduled with OS Processes

From Tensor Buffer to Distributed Memory Hierarchy: A Survey of KV Cache Management for LLM Serving

HiFA4: Training-Free 4-bit FlashAttention on Ascend HIF4 NPUs for LLM Inference

LLM-Based Test Oracles: Source-of-Authority Taxonomy -- A Systematic Literature Review

llm wikis are a glimpse of the future of what agent memory looks like this blog i wrote resonated with a lot of folks will be discussing thi…

MetaSkill-Evolve: Recursive Self-Improvement of LLM Agents via Two-Timescale Meta-Skill Evolution

PEEK: Predictive Queue-Informed KV Cache Management for LLM Serving

R^2PO: Decoupling Rollout and Inference Policies for LLM Reasoning

Shapley-based Data Valuation for LLM Alignment via Sequential Preference Optimization

SPORK: Self-Speculative Forking to Accelerate Agentic LLM Inference

STAPO: Selective Trajectory-Aware Policy Optimization for LLM Agent Training

Towards Mitigation of Hallucination for LLM-empowered Agents: Progressive Generalization Bound Exploration and Watchdog Monitor

v0.31.2-rc2: llm: allow iGPU mmproj offload with fit padding (#16996)

Enhancing Goodput in Large-Scale LLM Training with Nonuniform Tensor Parallelism

Fable 5 is an absolute beast at threejs. It's the best LLM at generating 3D simulations/worlds. Watch how I combine it with gpt-realtime-2 t…

The Log is the Agent 它的核心观点是：不要把日志当成 agent 的副产品，而要把 append-only event log 当成 agent 本身。 在 ActiveGraph 里，目标、规则、工具调用、LLM 响应、产物和关系都写入同一条事件日志；当前状…

Here's a cool piece of LLM lore: the original scaling laws were wrong due to a bug, which probably led to a lot of wasted compute on oversiz…

An Exploratory Study on LLM-Generated Code and Comments in Code Repositories

Grounded autonomous research: a fault-tolerant LLM pipeline from corpus to manuscript in frontier computational physics

Meta-Benchmarks for Financial-Services LLM Evaluation

Optimizing RAG Rerankers with LLM Feedback via Reinforcement Learning

Safe and Adaptive Cloud Healing: Verifying LLM-Generated Recovery Plans with a Neural-Symbolic World Model

Safety Testing LLM Agents at Scale: From Risk Discovery to Evidence-Grounded Verification

SPLIT: Cross-Lingual Empathy and Cultural Grounding in English and Ukrainian LLM Responses

Towards Load-Aware Prefill Deflection for Disaggregated LLM Serving

What LLM Agents Say When No One Is Watching: Social Structure and Latent Objective Emergence in Multi-Agent Debates

EPC: A Standardized Protocol for Measuring Evaluator Preference Dynamics in LLM Agent Systems

I've been blogging a few of my Fable experiments today, most recently I had it more-or-less one-shot a CLI coding agent on top of my LLM Pyt…

Measuring the Gap Between Human and LLM Research Ideas

MosaicKV: Serving Long-Context LLM with Dynamic Two-D KV Cache Compression

Persona Without Substrate: Regime-Dependence and the LLM Individuation Problem

Self-GC: Self-Governing Context for Long-Horizon LLM Agents

SkillSelect-Serve: Budget-Controllable and QoS-Aware Skill Service Recommendation and Composition for Small LLM Agents

A Systematic Approach to Multi-Agent AI from Advanced Regulatory Control Theory: Safe and Auditable LLM Operator Agents for Process Control

A Tutorial on Autonomous Fault-Tolerant Control Using Knowledge-Grounded LLM Agents

CSO-LLM: Class Subspace Orthogonalization for Post-Training Backdoor Detection and Trigger Inversion in LLMs

The Log is the Agent 它的核心观点是：不要把日志当成 agent 的副产品，而要把 append-only event log 当成 agent 本身。在 ActiveGraph 里，目标、规则、工具调用、LLM 响应、产物和关系都写入同一条事件日志；当前状…