AI News Digest 2026-11-03

特集

Anthropic、売上12倍も230億ドルの損失を計上
Anthropic grows revenue twelvefold but posts $23 billion loss

[01:55〜]

GoogleとMeta、Google TPU向けネイティブPyTorchサポート「TorchTPU」をオープンソース化
Google and Meta Open-source TorchTPU to Bring Native PyTorch Support to Google TPUs

[11:40〜]

AIコードレビューが1週間でLinuxのOverlayFSに潜む脆弱性を十数件発見 — 非特権ユーザーから悪用可能なバグも含むカーネルパッチが適用へ

[21:56〜]

話題コーナー

マスク氏、Grok Botの運用コストを「持続不可能」と発言し利用制限

[30:08〜]

Elon Musk throttles Grok Bot limits due to ‘unsustainably high’ costs — TechCrunch (AI)
SpaceXSI apparently cuts Grok Bot quotas weeks after Musk called it his fastest-growing product — The Decoder
Musk says Grok Bot cost unsustainably high — HN DeepSeek/Grok/OpenCode

どうしたんだ、Ubuntu
WTF Ubuntu?

[32:56〜]

AIツール紹介コーナー

DeepSeekが自律コーディングエージェント「dsh-superagent」をオープンソース公開

[35:10〜]

速報コーナー

Anthropic、Messages APIをPythonからRustへ全面移行　Rust経験者ゼロでClaudeが実装、CPU使用量40〜50%減

[43:03〜]

有料の新規GeminiプロジェクトはAPIリクエストを既定で保存

[44:03〜]

Claude Codeがバックグラウンドエージェントのトランスクリプトを定期送信していたと発覚

[44:49〜]

[アップデート] Kiro Crew v0.11.0 で Mate 2.0 が実装されカスタムクルーの挙動が変わりました

[45:58〜]

Dockerサンドボックスでコードはどこに置くべきか？
Where should your code live in a Docker Sandbox?

[47:02〜]

Uber、800以上のサーバーと5000以上のAIエージェントツールを管理するMCPゲートウェイを構築
Uber Builds MCP Gateway to Govern Access to 800+ Servers and 5,000+ AI Agent Tools

[47:47〜]

🔒 CLAUDE.md・AGENTS.mdに書いたルールをAIに守らせるには、どの行を残し、どこへ移し、何を消すべきか

[48:38〜]

週刊生成AI with AWS – 2026/10/26 週

[49:34〜]

[アップデート] Amazon Bedrock AgentCore Runtimeのセッション内で動くサーバーに直接アクセスできる機能Session endpointsが追加されました！

[50:23〜]

参考記事一覧を表示

DeepSeek Open-Sources dsh-superagent, a Router-Worker Coding Agent (importance: 79, dev_relevance: 85)
DeepSeek Boosts AI Agent Offering With New Creator Mode - Bloomberg.com (importance: 45, dev_relevance: 65)
Sorry Sam, my "quite terrible" home DRAM is at 100% now: patched Strata to make DeepSeek V4.1 agent sessions start 20× faster on a good old standard gaming GPU, no datacenter required (importance: 43, dev_relevance: 85)
Elon Musk throttles Grok Bot limits due to ‘unsustainably high’ costs (importance: 60, dev_relevance: 20)
Judge Says xAI Must Face NAACP Lawsuit On Data Center Pollution - Engadget (importance: 35, dev_relevance: 5)
Grok helps woman save $500/yr on meds - MoneyWise.com (importance: 15, dev_relevance: 5)
Hands-On With Grok Imagine Video: 13 Clips, Nine Tests and a $3 Bill for a Tech-News B-Roll Desk - Quasa.io (importance: 20, dev_relevance: 10)
OpenAI、ChatGPTとCodexの利用枠を共有へ ―2026年12月から、Instantモードは対象外 (importance: 0, dev_relevance: 67)
OpenAI・Google・Microsoft、robots.txtでAI学習の許可・拒否を一括指定する新たな仕組みに対応へ (importance: 0, dev_relevance: 37)
Anthropic clamps down on Claude’s AI video game clones (importance: 0, dev_relevance: 55)
GoogleのAI「Gemma 4」搭載衛星から「宇宙にいるAIからの世界初の応答」として俳句が送信される (importance: 41, dev_relevance: 15)
Claude Code sends each working background agent's full transcript to the model every 30 seconds to get a 3 to 5 word status line, and cost tools that read your transcripts never see those calls (importance: 48, dev_relevance: 90)
OpenAI might rebrand its data centers as ‘AI compute campuses’ (importance: 28, dev_relevance: 10)
Astra can read and write weights to improve Qwen (importance: 55, dev_relevance: 50)
Launch HN: Momentic (YC W24) – Agents that catch bugs by using your app (importance: 40, dev_relevance: 90)
Agentic Backdoors via Pretraining Poisoning (importance: 62, dev_relevance: 60)
Ask HN: Who is hiring? (November 2026) (importance: 5, dev_relevance: 25)
If you love a website, back it up (importance: 15, dev_relevance: 40)
Let Me Ch... (importance: 20, dev_relevance: 15)
Bitmovin (YC S15) Is Hiring Technical Graduates in Denver (Engineering/CS Focus) (importance: 5, dev_relevance: 30)
Egoless Mathematics in the AI Era (importance: 35, dev_relevance: 10)
WTF Ubuntu? (importance: 50, dev_relevance: 85)
We halved DNS build delay and cut memory use by 85% (importance: 40, dev_relevance: 90)
Moving our warehouse off DuckDB (importance: 45, dev_relevance: 85)
I Read the Clankers' Code (importance: 40, dev_relevance: 80)
Byesky: Wipe your Bluesky account clean (importance: 10, dev_relevance: 25)
My Favourite Laptop (importance: 5, dev_relevance: 20)
SOML – A Config Format for Humans (importance: 25, dev_relevance: 75)
Sneaking into the auditorium for the Windows 2000 launch rehearsal (importance: 10, dev_relevance: 35)
C64 Basic Dungeon Crawler: Stairs, Levels and Levelling Up (Part 11) (importance: 5, dev_relevance: 30)
Skill to Create Product Videos (importance: 35, dev_relevance: 70)
Show HN: I've updated my Linux System Call website and API (importance: 25, dev_relevance: 85)
The U.S. Census Bureau and "Super Intelligence" (importance: 43, dev_relevance: 18)
GPT-Rosalind-6 System Card Addendum (importance: 68, dev_relevance: 35)
Ask HN: Which TV do you recommend for privacy? (importance: 15, dev_relevance: 30)
Introducing the Grid Protectors Program (importance: 40, dev_relevance: 5)
LifeVal (importance: 30, dev_relevance: 15)
The latest AI news we announced in October 2026 (importance: 25, dev_relevance: 30)
Game Off 2026: The theme is SHIFT (importance: 10, dev_relevance: 35)
JetBrains Earns 180 G2 Badges in Fall 2026, Recognized for Enterprise Leadership, Usability, and Developer Experience (importance: 10, dev_relevance: 40)
Livestream: Scaling Agentic Development With Air Governance (importance: 30, dev_relevance: 70)
New Bug-Fix Releases for MPS Versions 2026.2 RC2, 2026.1.2, 2025.3.3, 2025.2.5, and 2025.1.5 (importance: 0, dev_relevance: 80)
The Ecosystem is the Answer: Docker, the Blueprint Alliance, and the Importance of Collaboration in the Agentic Era (importance: 40, dev_relevance: 60)
Where should your code live in a Docker Sandbox? (importance: 45, dev_relevance: 85)
Build World Models for Interactive Simulation with NVIDIA Cosmos (importance: 35, dev_relevance: 50)
How NVIDIA Factory Operations Blueprint Coordinates AI Agents to Investigate Production Issues (importance: 35, dev_relevance: 45)
Raising the quality of insights in agentic analytics (importance: 30, dev_relevance: 40)
Stop counting tokens, start counting what ships (importance: 35, dev_relevance: 60)
The agentic web is here (importance: 30, dev_relevance: 40)
Coming in 2027: Apple’s $1,000 home robot for FaceTime and switching on lights (importance: 35, dev_relevance: 5)
42AI comes out of stealth with $62.5M and a world model that can ‘predict human behavior’ (importance: 35, dev_relevance: 15)
Omen AI adds liquid know-how in $16 million Aquatic Labs acquisition (importance: 15, dev_relevance: 10)
2 days till TechCrunch Founder Summit: Boston founders don’t need a Silicon Valley address to build something big (importance: 5, dev_relevance: 10)
AI can make tech CEOs’ lives easier. What about everyone else? (importance: 15, dev_relevance: 10)
OpenAI pitches data center power plants as a way to protect nearby areas (importance: 40, dev_relevance: 5)
The New York Times sues Brave web browser for AI scraping and bypassing paywalls (importance: 45, dev_relevance: 20)
Google and Meta Open-source TorchTPU to Bring Native PyTorch Support to Google TPUs (importance: 65, dev_relevance: 90)
Uber Builds MCP Gateway to Govern Access to 800+ Servers and 5,000+ AI Agent Tools (importance: 45, dev_relevance: 85)
Article: Systemic Approach to Memory, Knowledge, and Context in Agentic AI Architectures (importance: 35, dev_relevance: 75)
AWS Capacity Largely Sold Out Through 2027, With On Demand Headroom Reserved (importance: 40, dev_relevance: 40)
Guarding the Gates of Knowledge (importance: 25, dev_relevance: 15)
Anthropic grows revenue twelvefold but posts $23 billion loss (importance: 78, dev_relevance: 10)
One brutal Minecraft night scared 50 AI agents into inventing chicken handoff protocols, a Fertilizer Officer, and fishing shifts (importance: 25, dev_relevance: 15)
Saudi Aramco takes 49 percent stake in AI firm Humain with $2.6 billion deal (importance: 45, dev_relevance: 5)
OpenAI prepares lobbying blitz for a post-midterm Washington (importance: 35, dev_relevance: 5)
Adobe is testing how well AI agents can use editing tools like a Photoshop professional would (importance: 40, dev_relevance: 50)
Empathetic prompts can turn AI chatbots into seemingly traumatized conversational partners (importance: 35, dev_relevance: 20)
Chinese AI models are politically complex, not just government mouthpieces, study finds (importance: 30, dev_relevance: 15)
OpenAI is steering ChatGPT users toward more paid usage (importance: 40, dev_relevance: 10)
Google wants video AI to understand what people need as they work (importance: 30, dev_relevance: 25)
10月26日〜11月2日、預かり鍵と未修正のカーネル・証明書 (importance: 0, dev_relevance: 80)
タッチ対応のMacBook Proが予約開始、M5 ProとM5 Maxで11月6日発売 (importance: 25, dev_relevance: 30)
AWS Private CAのKubernetes発行元に認可不備、更新前の確認が必要 (importance: 40, dev_relevance: 85)
Jepsen、ParticleDBに96件の不具合　新規導入の停止を勧告 (importance: 40, dev_relevance: 85)
同じ小規模アプリを12種のJSで作らせた費用、順位は揺れが大きい (importance: 45, dev_relevance: 85)
Chrome 157でHTMLをCanvas内に描けるAPIが安定版になった (importance: 45, dev_relevance: 85)
有料の新規GeminiプロジェクトはAPIリクエストを既定で保存 (importance: 50, dev_relevance: 85)
Chainalysis、AIエージェントを使った攻撃で社内システムに侵入され顧客の氏名・メールが流出　調査内容や認証情報の流出は確認されず (importance: 45, dev_relevance: 40)
Proton、CEOがペンテスト要望を「負け犬の発想」と退けた社内Slackが流出し謝罪　社内AIツールとレビュー手順の公開を約束 (importance: 35, dev_relevance: 40)
Anthropic、Messages APIをPythonからRustへ全面移行　Rust経験者ゼロでClaudeが実装、CPU使用量40〜50%減 (importance: 50, dev_relevance: 90)
Cisco Talos、盗んだ資格情報をLLMで順位付けするマルウェアを報告 (importance: 45, dev_relevance: 35)
Cloudflare、非公開IPのオリジンにWAFとWorkersを通す機能を一般提供 (importance: 40, dev_relevance: 75)
GitHub Actions、対象のLinuxジョブをAWSを含む複数クラウドで実行 (importance: 40, dev_relevance: 85)
GitHub、MarkdownからActionsを作るAgentic Workflowsを一般提供 (importance: 0, dev_relevance: 90)
ChatGPTの共有リンクで非公開会話を読めた欠陥、OpenAIが修正 (importance: 0, dev_relevance: 45)
NVIDIA、指示で体裁を変える音声モデルNemotron 3.5 Transcribeを公開 (importance: 45, dev_relevance: 40)
第135回　MySQL 26.10リリース、PostgreSQL Conference Japanの注目セッション、Tsurugiと分散量子コンピュータ (importance: 22, dev_relevance: 67)
Deep Principle Releases Hydrogen-1.0: A 753B Open-Weight Model for Scientific Discovery (importance: 40, dev_relevance: 30)
NVIDIA Releases Auto Ontology: An Open Semantic Layer That Grounds AI Agent SQL (importance: 40, dev_relevance: 85)
Meet qwenGPU: Qwen3.5-2B With 45 Compiled GPU Layers That Run CUDA (importance: 25, dev_relevance: 45)
Nace AI Open-Sources Drex Max v1: A 27B Decision Model That Runs Locally on Apple Silicon (importance: 0, dev_relevance: 40)
More Rules, Longer Lists? Benchmarking Self-Evolving Agents on a Hidden-Rule Curriculum (importance: 35, dev_relevance: 40)
Beyond Plausible Prediction: Safety Sufficiency in Web World Models (importance: 30, dev_relevance: 25)
Circumventing the Computational Hardness of Computational Arbitrage (importance: 25, dev_relevance: 10)
Incongruence-Aware Evaluation of Multimodal Speech Emotion Recognition in Audio Large Language Models (importance: 25, dev_relevance: 15)
ChatGPT as a Surveillance Assistant? An Exploratory Case Study of AI-Assisted Surveillance Risks (importance: 35, dev_relevance: 15)
Does Recursive Knowledge Construction Help Agents Verify Consensus Claims in Science? (importance: 30, dev_relevance: 35)
Towards Portable Specification of the Agent Harness (importance: 40, dev_relevance: 65)
Agentic World Modeling for Agent Test-time Scaling without Environment Replica (importance: 30, dev_relevance: 50)
DiRAFT: Privacy-Aware Financial Transcript Synthesis from Aggregate Discourse Priors (importance: 20, dev_relevance: 25)
The First Few Tokens Matter: Understanding Chain-of-Thought Controllability (importance: 35, dev_relevance: 25)
Benchmarking Forecasting Agents on Yesterday's Tomorrows (importance: 25, dev_relevance: 25)
Opening the Box Changes the Cat: LLMs Detect Chain-of-Thought Monitoring - What now? (importance: 30, dev_relevance: 20)
P-ContextBench: From Personal Fact Recall to Personalized Agent Behavior (importance: 25, dev_relevance: 25)
Nedimo: An Evidence-Grounded Multilingual Pipeline for AI-Generated Sleep Stories (importance: 20, dev_relevance: 20)
The psychophysics of affect in large language models (importance: 25, dev_relevance: 15)
Can Personal Requirements Become Supervision for Planning? (importance: 20, dev_relevance: 20)
Violet Teaming in Practice: Lessons from a Pilot Expert Workshop on Governing Complex AI Systems (importance: 25, dev_relevance: 10)
ReplayRCA: Measuring Post-hoc Diagnosis and Live Oversight of Agent Failures by Counterfactual Replay (importance: 35, dev_relevance: 55)
Capability-Grounded Agentic Planning (importance: 35, dev_relevance: 55)
Consensus Within, Competition Between: Decentralized LLM Teams for Perimeter Defense (importance: 20, dev_relevance: 20)
LanG 2.0: Energy-Efficient Agentic AI for Semi-Automated Security Operations at Scale (importance: 30, dev_relevance: 40)
BABA-Bench: Evaluating Long-Horizon Patient-Advocacy Agents (importance: 25, dev_relevance: 20)
Optimization Techniques for Solving the Minimum Red--Blue Separating Set Problem with an Application to Human--AI Text Separation (importance: 15, dev_relevance: 25)
Payroll Fraud Detection in Synthetic HR/ERP Graphs: An Inductive Comparison of Rules, Path Features, and Graph Neural Networks (importance: 20, dev_relevance: 25)
CIRCE Divines LLM Task Complexity from Constraint Structure (importance: 35, dev_relevance: 35)
Verifying Text-to-SQL Queries under Changing Business Semantics (importance: 35, dev_relevance: 55)
QUAL: Quantile Alignment Improves Zero-Shot Time-Series Foundation Models (importance: 30, dev_relevance: 40)
Differential rank preservation manifold learning (importance: 20, dev_relevance: 30)
Calibration of Large Model Quantization based on HiFloat4 Format (importance: 30, dev_relevance: 50)
Predicting Human-Gain Curves under Local Proxy Optimization from Repeated Ratings (importance: 30, dev_relevance: 35)
Confident Fools or Careful Searchers? A Behavioural Study of When LLMs Search Beyond Their Static Internal Memory (importance: 40, dev_relevance: 55)
When Should AI Agents Ask? Clarification Requests Under Limited Human Attention (importance: 35, dev_relevance: 55)
Counterfactual Pairwise Augmentation Improves Small Pointwise LLM Judges (importance: 30, dev_relevance: 50)
Lifelong Learning of Scheduling Rules with Surrogate-assisted Genetic Programming for Dynamic Flexible Job Shop Scheduling (importance: 10, dev_relevance: 10)
FrontierMat: Benchmarking Language Models at the Frontier of Materials R&D (importance: 35, dev_relevance: 15)
LLM Detoxification via Transition-Aware Reward-Guided Representation Sampling (importance: 25, dev_relevance: 35)
ImplicitHop: Evaluating knowledge integration across context and memory (importance: 25, dev_relevance: 20)
Researcher-in-the-Loop Evolutionary Discovery of Entanglement-Assisted Quantum Code Families (importance: 30, dev_relevance: 20)
RTL-Interact: Training EDA-Interactive Models for Multi-Task RTL Design (importance: 30, dev_relevance: 60)
TRIDENT and SHAKTI: A Framework and Benchmark for Hindi AI-Text Detection (importance: 20, dev_relevance: 15)
Experts Exhibit Structured Redundancy: Structure-Preserving Expert Pruning for Mixture-of-Experts Large Language Models (importance: 35, dev_relevance: 45)
MEMIG: Self-evolving Memory Construction for Antibody Optimization (importance: 20, dev_relevance: 25)
Rome Wasn't Built in One Turn: Replay-Grounded Reward Signals for Long-Horizon Freeciv Agents (importance: 20, dev_relevance: 25)
From 1D Benchmarks to Complex 3D Geometries: Can Agents Reliably Specify Problems for Automated PDE Simulation? (importance: 25, dev_relevance: 40)
SciGraphSelect: Auditable Scientific Data Selection through Knowledge--Reasoning Graphs (importance: 20, dev_relevance: 20)
SAT-ACT: Learning CDCL Branching from Outcome-Derived Action Preferences (importance: 25, dev_relevance: 35)
Self-Improvement with Auxiliary Rewards: Learning from Verified Traces and Evaluating Transfer (importance: 20, dev_relevance: 25)
Subliminal Reinforcement Learning: traits travel through a judge's scores (importance: 40, dev_relevance: 35)
SERIAL-7: Deriving, Realizing, and Preserving Machine-Verifiable Outputs Across Representations (importance: 25, dev_relevance: 50)
Operon-S1: a fast decision controller for long-horizon biomedical agents (importance: 25, dev_relevance: 35)
Do Transaction Prices Reflect Visual Preference? A Diagnostic Study with Multimodal Language Models (importance: 15, dev_relevance: 10)
LEAP: Efficient LLM Inference Scheduling with Generalizable Length Prediction (importance: 35, dev_relevance: 70)
RECITE: Revising Eligibility Criteria Incorporating Textual Evidence (importance: 20, dev_relevance: 10)
From the Zoo to the BULLPEN: Sample-Efficient LLM Profiling via Multi-Channel Model Embeddings (importance: 25, dev_relevance: 35)
Ask on Demand: Targeted Peer Communication for Parallel Agentic Systems (importance: 30, dev_relevance: 65)
Online Evidence-Containment Prediction over Text Spans under Distinct Queries (importance: 15, dev_relevance: 25)
Cross-Interface Authorization for AI Computer-Use Agents (importance: 35, dev_relevance: 55)
scAnyOmics: Foundation-Model Distillation for Any-to-Any Completion of Multi-omics Virtual Cells (importance: 20, dev_relevance: 10)
CryoAtlas: A Large Curated Dataset and Unified Benchmark for Cryo-EM Atomic Model Building (importance: 15, dev_relevance: 10)
Evidence-Guided Recovery for AI Agents with Ambiguous Tool Outcomes and Partial Execution (importance: 35, dev_relevance: 65)
EvoMCP: Co-evolving MCP agents and training data (importance: 25, dev_relevance: 45)
Multi-Test Nation Bias Benchmarking of LLMs with Explicit Unbiased Baselines (importance: 20, dev_relevance: 10)
Specification-Guided Generation for Large Language Models (importance: 25, dev_relevance: 50)
Verify--Backtrack Loop: Structured Precondition Recovery for Long-Horizon Language Agents (importance: 25, dev_relevance: 50)
Connectivity-Aware Client Selection in Decentralized Federated Learning (importance: 10, dev_relevance: 20)
Patch-Based Deep Kernel Learning for Data Assimilation and Bias Correction with Extremely Sparse Observations (importance: 15, dev_relevance: 10)
Real Computation or Merely Naming? A Pre-Registered Mechanism Audit of 25 Cognitive-Architecture Modules in an LLM-Based Agent System (importance: 30, dev_relevance: 50)
What Should LLMs Forget: Reinforcement Unlearning of Knowledge Associations (importance: 20, dev_relevance: 20)
Chance-Constrained Inspection Viewpoint Sequencing for On-Orbit Servicing via Safe Reinforcement Learning (importance: 10, dev_relevance: 15)
Risk-Sensitive Mission Planning for Multi-Target On-Orbit Inspection via Distributional Reinforcement Learning (importance: 10, dev_relevance: 15)
Breaking Information Islands in Sparse Tuning via Small-World Connectivity (importance: 20, dev_relevance: 35)
Radial Kernel Attention for Long-Range Modeling in Polymer Machine Learning Force Fields (importance: 10, dev_relevance: 15)
TSBPO: Tree-Structured Branching for Policy Optimization in Interactive Language Agents (importance: 25, dev_relevance: 50)
Advancing Clinical Trial Outcome Prediction with Domain-Specific Large Language Models (importance: 15, dev_relevance: 10)
Multimodal In-Context Learning with Explicit Rule Induction (importance: 15, dev_relevance: 15)
Clinical-Protocol-Aware Cyber Resilience for Connected Clinical Technologies Under AI-Enabled Adversarial Conditions (importance: 25, dev_relevance: 25)
Whose Values, Which Contexts, and What Signals? Grounding Cultural Value Alignment with Re3Culture (importance: 20, dev_relevance: 10)
RECAST: REconstruction of Chain-of-Thought via Auditable Structured Traces (importance: 30, dev_relevance: 40)
ReuScope: Auditing Data Reuse and Evaluation Sensitivity in Capsule Vision 2024 (importance: 10, dev_relevance: 20)
Dynamical Memory Transformer: Selective Recurrent Memories with Constant Cost Attention (importance: 25, dev_relevance: 35)
Learning to Interact and Direct: Co-Evolving Video Director Agents (importance: 15, dev_relevance: 15)
Fuzzy selective state model for forecasting the dynamics of consumer preferences (importance: 10, dev_relevance: 15)
Automated Keyword Generation Using LLMs in a High-Volume Data Virtualization Service (importance: 25, dev_relevance: 50)
Using Machine Learning to Measure Self-Efficacy Belief Through Learners' Interaction Traces in Technology-Enhanced Learning (importance: 10, dev_relevance: 10)
ReX-GG: a LLM Ensemble Pipeline for Relation-extraction and Graph Generation (importance: 20, dev_relevance: 40)
Structure-Aware Skill Retrieval and Arbitration for Optimization Modeling (importance: 20, dev_relevance: 35)
Sensorium: An Open-Source Multimodal Data Collection and Curation Library for Embodied Surgical AI (importance: 20, dev_relevance: 45)
Optimizing Stochastic Policy Geometry for Few-Step Flow Reinforcement Learning (importance: 15, dev_relevance: 25)
Prune Gradually, Adapt Sparsely: Efficient Pruning of Large Language Models at Extreme Sparsity (importance: 30, dev_relevance: 40)
When Does Test-Time Search Help World Action Models? An Empirical Audit of Candidate Selection and Simulator Integrity (importance: 15, dev_relevance: 20)
ID2Pattern: Pretraining Generalizable Identity Behavior Patterns for Transaction Risk Foundation Models (importance: 15, dev_relevance: 20)
DAMI: Dual-Axis Memory Initialization for Enhanced Long-Context Retrieval in Gated Linear Attention (importance: 25, dev_relevance: 30)
TradeWar: A Benchmark for Execution-Grounded Trade Negotiation with Adaptive LLM Agents (importance: 15, dev_relevance: 15)
Same Question, Different Answers: Auditing Persona-Dependent Variation in Six LLMs on Smoking-Cessation Advice (importance: 30, dev_relevance: 10)
The Knowledge Layer Model (KLM): Evidence-Visible AI Output, Routing Diagnostics, and Negative Results (importance: 20, dev_relevance: 40)
CausalWAM: Learning the Causes of Actions in World Action Models (importance: 15, dev_relevance: 15)
ACCT: Asymmetric Cross-LLM Context Transfer (importance: 25, dev_relevance: 60)
Latent Sparse Attention: Towards Subquadratic Ultra-Long Context Attention (importance: 25, dev_relevance: 45)
Hear No Evil: Detecting Gradient Leakage by Malicious Servers in Federated Learning (importance: 15, dev_relevance: 15)
Leaner Training, Lower Leakage: Revisiting Memorization in LLM Fine-Tuning with LoRA (importance: 25, dev_relevance: 35)
Recycle only when you have to! (importance: 10, dev_relevance: 15)
CausalGrade: Evidence-Guided Causal Graph Scoring with Jev (importance: 10, dev_relevance: 20)
PrivacyFlow: Privacy in AI Agents through Information Flow Analysis and Consequence-Aware Risk Management (importance: 30, dev_relevance: 40)
Characterizing LLM Agent Behaviors in Module Shadowing Environments (importance: 30, dev_relevance: 55)
MetaVul: Agentic Harness Discovery for Vulnerability Detection (importance: 30, dev_relevance: 55)
Bayesian Distribution System State Estimation with Flow Matching (importance: 10, dev_relevance: 10)
RiskyPy: Toward Model-Calibrated Prompt Generation for Joint Security and Functionality Evaluation of LLM-Generated Code (importance: 25, dev_relevance: 60)
CLIC: Closed-Loop Intent Consistency for Securing LLM-Driven Control in Open RAN (importance: 20, dev_relevance: 20)
Safin-1.5: Safety-Native Architecture and Verifiable Reasoning for Recursive Safety Improvement (importance: 25, dev_relevance: 15)
"Shall We Approve?" Consequence-Aware Approval for Autonomous Coding Agents (importance: 40, dev_relevance: 60)
AutoHebb: Self-Modifying Neural Networks that Zero-Shot Infer and Continually Learn (importance: 15, dev_relevance: 15)
TAPE: Learning General Audio Effect Representations (importance: 15, dev_relevance: 20)
From Teleoperation to Long-Horizon Dexterity: Evaluating Visuomotor Skills and Offline Proxies for Dexterous Manipulation (importance: 15, dev_relevance: 15)
The Generative Gap Between Speculative Drafter and Continuous Diffusion Language Model (importance: 20, dev_relevance: 40)
GraphTFactor: A graph-based approach for proteins transcription factor prediction (importance: 5, dev_relevance: 10)
TSFLab: A Self-Improving Agent-Agnostic Workbench for Time Series Forecasting Research (importance: 25, dev_relevance: 50)
A Theoretical Framework for Understanding and Building "Any-task inference" Capable Foundation Models (importance: 15, dev_relevance: 15)
Kernel-based Interpolatory Methods for Context Compression of Attention Mechanisms (importance: 20, dev_relevance: 40)
On the Origin and Evolution of Training-Inference Mismatch in LLM Reinforcement Learning (importance: 35, dev_relevance: 55)
CAMEL: Cross-Architecture Masked Embedding Learning for Ultra-Lightweight Retinal Foundation Models (importance: 10, dev_relevance: 10)
Steering Vectors as a Training Signal in LLM Post-Training (importance: 25, dev_relevance: 35)
STAR: Steering Action Representations during Inference (importance: 15, dev_relevance: 15)
Towards Understanding Score-Seeking in LLMs (importance: 40, dev_relevance: 40)
Do Language Models Know What Compelling Scientific Evidence Looks Like? (importance: 15, dev_relevance: 10)
Dreams Are Cheaper Than Flights: Curated World-Model Data Improves Aerial Detection (importance: 30, dev_relevance: 35)
WebOverlay: High-Throughput Vision-Only Distillation for Extreme Aspect-Ratio Web Overlay Detection (importance: 30, dev_relevance: 45)
Characterizing Recoverable Expert-Utilization Signatures in Fine-Tuned Mixture-of-Experts LLMs (importance: 40, dev_relevance: 40)
Accept Changes, Reject Attacks: Robust Memory Management for LLM Agents (importance: 42, dev_relevance: 50)
bf16 Rounding Silently Discards Most RL Policy Updates: A Hidden Training-Inference Mismatch and an Unbiased One-Line Fix (importance: 48, dev_relevance: 65)
Where, When, and How Many Bits? Scaling Laws for SGD with Stochastic Rounding (importance: 25, dev_relevance: 35)
MUON-RT Reusing the Teacher's Trajectory for Momentum-Free Muon for LLM Self-Distillation (importance: 35, dev_relevance: 55)
One Verdict Is Not Enough: Regression Testing Learned World Models and Their Test Harnesses (importance: 30, dev_relevance: 60)
When Disclosure Is Compulsory: Evaluating Ride-Hailing Rebalancing Policies with Self-Interested LLM Driver Agents (importance: 25, dev_relevance: 15)
Lineage Is Memory: Parent Inheritance and Feedback Disclosure in Short-Budget LLM-Guided Optimization (importance: 35, dev_relevance: 60)
ARK: Avoiding Routing Collisions for KV Cache Transfer in Disaggregated LLM Inference (importance: 30, dev_relevance: 65)
ROSTrans: Empirically Benchmarking LLMs for ROS1-to-ROS2 Migration (importance: 30, dev_relevance: 70)
Context as Infrastructure: Knowledge Compilation for Retrieval in Agentic Systems with Harnesses (importance: 40, dev_relevance: 70)
Beyond CLIP: Calibrated (Few-Shot) Defect Detection With System One Decision Models (importance: 25, dev_relevance: 30)
From Evaluation to Evolution: Benchmarking Image Editing and Evolving Image Models (importance: 25, dev_relevance: 10)
Coarse3D: Controllable and Faithful Sparse Structure Generation for Image-to-3D (importance: 25, dev_relevance: 15)
ANGLER: Verifier-Grounded Benchmarking with an Adaptive Attacker for Web Agents (importance: 35, dev_relevance: 55)
In-Model Context Compression with Agentic Soft Tokens for Long-Horizon LLM Agents (importance: 35, dev_relevance: 55)
ART of PIV: Agentic Real-Time Particle Image Velocimetry for Active Flow Control (importance: 20, dev_relevance: 25)
Erase the Injected Region: Multimodal Prompt-Injection Defense with On-Policy Self-Distillation (importance: 40, dev_relevance: 50)
GenDSR: Transferring Representations from Video Generation Models for Dynamic Spatial Reasoning (importance: 25, dev_relevance: 15)
Transferring Peptide-Target Interaction Patterns to Small Molecules by Joint Guidance for Diffusion and Flow Matching (importance: 20, dev_relevance: 15)
With foundation-model features, aggregator complexity does not improve cross-hospital discrimination in gastric whole-slide image classification (importance: 20, dev_relevance: 15)
Pinching Antennas for ISAC: Waveguide Architectures, Solutions, and Future Directions (importance: 0, dev_relevance: 20)
From prediction error to sparse events: Biologically inspired signed-dyadic coding for Mandarin keyword spotting (importance: 20, dev_relevance: 30)
ProSearch: Benchmarking Multi-Constraint Protocol Retrieval in Experimental Science (importance: 25, dev_relevance: 35)
Distilling Knowledge from Search: Decoder Model Factory for Quantum LDPC Codes (importance: 20, dev_relevance: 20)
When Safety Checks Become Privacy Risks: A Privacy-Preserving Paradigm for LLM Agent Safety Auditing (importance: 35, dev_relevance: 55)
SparseGDN: Chunk Sparse in Linear Attention (importance: 30, dev_relevance: 50)
BlindToChange: AI Agents Can Find Changes When Asked But Do They Look When Needed? (importance: 35, dev_relevance: 40)
From Representation to Readout: Mechanistic Analysis of Membership Signals in Language Models (importance: 25, dev_relevance: 25)
KEEPER: Learning Control-Sufficient Textual Memory for Long-Horizon LLM Agents (importance: 30, dev_relevance: 50)
Sparse Text Diffusion under Dual Budgets: Context Connectivity and Expert Activation (importance: 25, dev_relevance: 25)
ORION: An Outcome-Driven LLM Advisor for RL-Based Multi-Domain 6G Slice Placement (importance: 20, dev_relevance: 20)
VLA Policies Do Not Know When Not to Act: Unintended Task Execution and Start Suppression (importance: 35, dev_relevance: 25)
Attention Maps Know When to Think: Training-Free Difficulty-Aware Chain-of-Thought for Reasoning Segmentation (importance: 25, dev_relevance: 20)
GRIT: Gradient Masking for Visually Misgrounded Tokens in Text-Rich Reasoning (importance: 25, dev_relevance: 25)
When Steering Shifts Discourse: Identifying and Intervening on Discourse-Level Changes due to Activation Steering in LLMs (importance: 30, dev_relevance: 30)
OT-CAL: Optimal-Transport-Guided Adaptive Sample Selection for Efficient Visual Learning (importance: 25, dev_relevance: 35)
When the Past Shapes Action: Understanding Embodied Memory (importance: 25, dev_relevance: 15)
Simple Features Beat Most AI Control Monitors (importance: 35, dev_relevance: 45)
Latent-Lens: Visual Perception in Small Language Models Through Latent Communication (importance: 25, dev_relevance: 25)
Different Models, Shared Failures: Auditing Frozen-Encoder Pools for Medical Segmentation (importance: 20, dev_relevance: 15)
Implicit Assumptions About Tasks in Continual Learning (importance: 20, dev_relevance: 15)
Accurate and Auditable ECG Classification via Visibility Graph Neural Networks (importance: 20, dev_relevance: 10)
Backpropagation-free Deep Learning for Time-Series Classification (importance: 25, dev_relevance: 30)
SimXAI: A Simple Drop-in Layer Increases Interaction Quality and Understanding in Conversational XAI (importance: 25, dev_relevance: 40)
AI evaluators favor job applications written by certain AI models, independent of candidate quality (importance: 40, dev_relevance: 25)
General-purpose foundation models surpass specialized ones on chemical property prediction tasks (importance: 25, dev_relevance: 20)
Observable-Specific Validity Tests for Scientific AI: Controlled Approximations and Intervention Response (importance: 25, dev_relevance: 20)
Aligning MLLMs with the Latent Structure of Human Cognition via Behavior-Derived Semantic Dimensions (importance: 25, dev_relevance: 10)
Chronos-2: A foundation model for multivariate time series forecasting (importance: 50, dev_relevance: 45)
SecPO: Principled Adversarial Training for Prompt Injection Security (importance: 35, dev_relevance: 40)
AutoAnalysis: An automated, self-updating meta-analysis of human-AI collaboration (importance: 25, dev_relevance: 20)
Know When the Equation Is Wrong: Structural Uncertainty for LLM-Guided Symbolic Regression under Extrapolation (importance: 25, dev_relevance: 15)
There Is No Free Loop: When Does Looping a Trained Model Pay for Its Inference Compute? (importance: 35, dev_relevance: 40)
Learning Clean Generative Processes from Noisy Measurements (importance: 25, dev_relevance: 10)
Near-Optimal Second-Order Convex Finite-Sum Optimization under Individual Smoothness (importance: 5, dev_relevance: 15)
CRoS: Causality Reward on Scrambled CoTs Improves Accuracy and Reasoning Faithfulness (importance: 30, dev_relevance: 30)
Subquadratic Attention with a Hierarchical Tree-Structured Indexer (importance: 30, dev_relevance: 40)
RoboArchitect: Automating the Design of Robotic Agents with Persistent Memory (importance: 30, dev_relevance: 50)
Hierarchical World Models with Implicit Dynamics (importance: 25, dev_relevance: 15)
Shared Modular Recurrence in Contextual MDPs for Universal Morphology Control (importance: 15, dev_relevance: 10)
A Hybrid SMT-NRA Solver: Integrating 2D Cell-Jump-Based Local Search, MCSAT and OpenCAD (importance: 5, dev_relevance: 35)
Ax-Prover: A Deep Reasoning Agentic Framework for Theorem Proving in Mathematics and Quantum Physics (importance: 25, dev_relevance: 35)
ThinkSafe: Self-Generated Safety Alignment for Reasoning Models (importance: 30, dev_relevance: 30)
SoK: Agentic Retrieval-Augmented Generation (RAG): Taxonomy, Architectures, Evaluation, and Research Directions (importance: 35, dev_relevance: 50)
Controllable User Simulation (importance: 20, dev_relevance: 25)
Teaching Large Language Models When Not to Know: Learning Temporal Critique for Ex-Ante Reasoning (importance: 25, dev_relevance: 20)
Generative Recursive Reasoning (importance: 25, dev_relevance: 15)
GENSTRAT: Toward a Science of Strategic Reasoning in Large Language Models (importance: 20, dev_relevance: 15)
Physics-Audited Agentic Discovery in Scientific Machine Learning (importance: 25, dev_relevance: 35)
Neuro-Symbolic Meta-Policies for Temporal Knowledge-Graph Memory under Partial Observability (importance: 20, dev_relevance: 25)
FedPA-LoRA: Product-Aligned Framework for Mitigating Aggregation and Initialization Errors in Heterogeneous Federated LoRA (importance: 25, dev_relevance: 40)
From Association to Causation: Improving Retrieval Precision of Retrieval-Augmented Generation via Causal Relations and an Attention Mechanism (importance: 25, dev_relevance: 40)
SIRF: A Spec-Internalized Risk Foundation Model for Industrial Content Risk Control (importance: 25, dev_relevance: 20)
A Unified Per-Token Gating Family for On-Policy Distillation: FKL/RKL Mixing with Multi-Channel and Bias Coefficients (importance: 25, dev_relevance: 35)
Unified Agentic Video Editing Across Levels of Complexity and Creativity (importance: 20, dev_relevance: 15)
LazyAgent: Demand-Driven Materialization and Physical Optimization of Agentic Programs (importance: 35, dev_relevance: 70)
Fine-Tuning VLM for Enhancing AI's Spatial Intelligence: Understanding 3D and 2D Rotations (importance: 20, dev_relevance: 10)
Verification and Self-Improvement in Agentic AI: Foundations and Limits (importance: 25, dev_relevance: 40)
From Stages to States: Unified Iterative Reasoning for MLLMs (importance: 20, dev_relevance: 15)
WeFair: Fidelity-Aware Alignment from Human Feedback for Personalized Query Rewriting in Short-Video Search (importance: 20, dev_relevance: 20)
Does a Governed Confidence Score Know When a Text-to-SQL Answer Is Right? An Exploratory Study on BIRD and a Pre-registered Confirmation on Spider (importance: 30, dev_relevance: 55)
Terminal-RSI: Recursive Self-Improving Terminal Agents via Recursive Self-Rewrite and Policy-Refine (importance: 35, dev_relevance: 55)
Self-Building Labs (importance: 25, dev_relevance: 15)
From Prediction to Action: Population Health World Model for Digital Medicine (importance: 20, dev_relevance: 10)
Model Organisms of Collusion: Corrective Transfer in Untrusted Monitors (importance: 35, dev_relevance: 40)
Reject, Remember, Retry: Why Multi-Turn Attackers Should Remember Their Mistakes (importance: 25, dev_relevance: 20)
AgentOrch: How Close Can Global Orchestration Come to Step-by-Step Interaction? (importance: 30, dev_relevance: 60)
LegoSR: A Unified Framework for Self-improving Agent-Guided Scientific Discovery (importance: 20, dev_relevance: 35)
Logic Hijacking of RAG: Single-Document Poisoning via Local Optimization (importance: 30, dev_relevance: 45)
MuSID-KD: Multigranularity Knowledge Distillation for Generative Recommendation with Semantic IDs (importance: 20, dev_relevance: 25)
SpectRAG: Weighted Chunk Graphs for Token-Budgeted Context Selection in Multi-Hop Retrieval-Augmented Generation (importance: 25, dev_relevance: 40)
Concordat: Typed Sessions for LLM Agents (importance: 30, dev_relevance: 65)
Learning to Reason from Reference Samples Alone: Reinforced Adversarial Learning for Modern LLM Post-Training (importance: 25, dev_relevance: 20)
What Breaks When Facts Become Stories? The Distortions Source-Agreement Metrics Miss When Models Retell Real Events (importance: 20, dev_relevance: 15)
MoLRSI: A Recursive Self-Improving AI agent for Molecular Discovery (importance: 25, dev_relevance: 30)
EvoTeam: Multi-Agent Collaboration via Evolving Teams (importance: 30, dev_relevance: 45)
Algebraic World Model: From Exploration to Proofs (importance: 25, dev_relevance: 15)
Time Will Tell: Harness Time to Denoise Collaborative Filtering (importance: 25, dev_relevance: 30)
Integrating AI for Human-Centric Breast Cancer Diagnostics: A Multi-Scale and Multi-View Swin Transformer Framework (importance: 25, dev_relevance: 15)
Beyond Public Access in LLM Pre-Training Data (importance: 45, dev_relevance: 20)
Optimizing Anytime Reasoning via Budget Relative Policy Optimization (importance: 35, dev_relevance: 30)
LLINBO: Trustworthy LLM-in-the-Loop Bayesian Optimization (importance: 30, dev_relevance: 25)
Fully Offline Reinforcement Learning (importance: 25, dev_relevance: 20)
VisualSphinx: Large-Scale Synthetic Vision Logic Puzzles for RL (importance: 35, dev_relevance: 20)
PPS: Plug-and-Play Saccadic Vision for Fine-Grained Classification (importance: 25, dev_relevance: 20)
MLLMEraser: Achieving Test-Time Unlearning in Multimodal Large Language Models through Activation Steering (importance: 30, dev_relevance: 15)
On the Faithfulness of Visual Thinking: Measurement and Enhancement (importance: 30, dev_relevance: 15)
Speech-DRAME: A Framework for Human-Aligned Benchmarks in Speech Role-Play (importance: 25, dev_relevance: 20)
DiTGaze: Geometry-Conditioned Diffusion Augmentation for Few-Shot Gaze Estimation (importance: 20, dev_relevance: 15)
Instruction-Tuning Open-Weight Language Models for BPMN Model Generation (importance: 30, dev_relevance: 50)
Scaling Causal Mediation for Complex Systems: A Framework for Root Cause Analysis (importance: 20, dev_relevance: 35)
Variational decomposition autoencoding improves disentanglement of latent representations (importance: 20, dev_relevance: 10)
TIRx: A Unified Tile Primitive ML Compiler Across Heterogeneous Backends (importance: 35, dev_relevance: 75)
Eroding the Truth-Default: A Causal Analysis of Human Susceptibility to Foundation Model Hallucinations and Disinformation in the Wild (importance: 25, dev_relevance: 10)
Safe Reinforcement Learning via Recovery-based Shielding with Gaussian Process Dynamics Models (importance: 20, dev_relevance: 15)
Interpretability-by-Design with Accurate Locally Additive Models and Conditional Feature Effects (importance: 20, dev_relevance: 25)
Place-it-R1: Unlocking Environment-aware Reasoning Potential of MLLM for Video Object Insertion (importance: 25, dev_relevance: 10)
SUREON: A Benchmark and Vision-Language Model for Surgical Reasoning (importance: 25, dev_relevance: 5)
Proof-Carrying Materials: Falsifiable Safety Certificates for Machine-Learned Interatomic Potentials (importance: 25, dev_relevance: 25)
Open Biomedical Knowledge Graphs at Scale: Construction, Federation, and AI Agent Access with Samyama Graph Database (importance: 25, dev_relevance: 50)
Specification-Aware Distribution Shaping for Robotics Foundation Models (importance: 20, dev_relevance: 15)
Enhancing Safety of Large Language Models via Embedding Space Separation (importance: 25, dev_relevance: 15)
STRIATUM-CTF: A Protocol-Driven Agentic Framework for Autonomous CTF Solving (importance: 25, dev_relevance: 40)
Can Humans Tell? A Dual-Axis Study of Human Perception of LLM-Generated News (importance: 25, dev_relevance: 5)
M$^\star$: Every Task Deserves Its Own Memory Harness (importance: 35, dev_relevance: 50)
A Survey on Long-Term Memory Security in LLM Agents: Attacks, Defenses, and Governance Across the Memory Lifecycle (importance: 35, dev_relevance: 40)
Autoregressive Diffusion for Compositional Text-to-Scene Generation (importance: 15, dev_relevance: 10)
How Fast Should a Model Commit to Supervision? Training Reasoning Models on the Tsallis Loss Continuum (importance: 25, dev_relevance: 20)
AMMA: A Multi-Chiplet Memory-Centric Architecture for Low-Latency 1M Context Attention Serving (importance: 30, dev_relevance: 25)
Sparkle: Realizing Lively Instruction-Guided Video Background Replacement via Decoupled Guidance (importance: 20, dev_relevance: 10)
EvoMemBench: Benchmarking Agent Memory from a Self-Evolving Perspective (importance: 30, dev_relevance: 40)
DRScaffold: Boosting Dense-Scene Reasoning in Lightweight Vision Language Models (importance: 20, dev_relevance: 15)
A Relative-Computability Theory of Self-Improving Agents (importance: 15, dev_relevance: 15)
Semantic Optimal Transport for Sparse Autoencoder Feature Matching and Circuit Compression (importance: 20, dev_relevance: 20)
Analyzing the Optimization Conflict Between Image- and Text-Based Person Re-Identification (importance: 15, dev_relevance: 10)
How Do Video Foundation Models Encode Intuitive Physics? Probing Across Pretraining Paradigms (importance: 20, dev_relevance: 10)
Beyond IID: How General Are Tabular Foundation Models, Really? (importance: 25, dev_relevance: 45)
Towards Diverse and Comprehensive Benchmarks for Mutual Information Estimation (importance: 15, dev_relevance: 20)
DAUPNet: Domain-Aware Uncertainty Modeling for Reliable Prototype Discrimination in Cross-Domain Few-Shot Semantic Segmentation (importance: 15, dev_relevance: 10)
Free energy landscape of Dense Associative Memory (importance: 15, dev_relevance: 10)
CogBattery: Human-Referenced Evaluation of Cognitive Task Behavior and Structure in Language Models (importance: 25, dev_relevance: 20)
ReBRAC-v2: The Return of the King (importance: 25, dev_relevance: 25)
GUI-Lens: Enabling Precise GUI Grounding for Off-the-Shelf VLMs via Coarse-to-Fine Cropping (importance: 25, dev_relevance: 40)
Simulation-free Unbalanced Dynamic Optimal Transport with General Growth Penalty (importance: 10, dev_relevance: 15)
TamilEOT: A Dataset and Model for Semantic End-of-Turn Detection in Tamil Telephone Speech (importance: 20, dev_relevance: 35)
HyCO: A Hybrid Neural Solver for Combinatorial Optimization (importance: 15, dev_relevance: 15)
Learning and Transferring Closed-Loop Robot Software (importance: 30, dev_relevance: 55)
A Comprehensive View of Fairness through Distributional Stability (importance: 10, dev_relevance: 10)
Detect and Suppress: A Mechanistic Defense against Adversarial Patches in VLA Models (importance: 25, dev_relevance: 15)
What Does a Harness Repair? Three Preregistered Studies Around a Longevity-Tuned Language Model (importance: 25, dev_relevance: 40)
Vision Transformer Ensembles for Panoramic Street Segmentation (importance: 10, dev_relevance: 10)
Learning to Orchestrate Evolutionary Search: Progression-Aware Deep Reinforcement Learning for Dynamic DE-CMA-ES Coordination in Optimization and Structural Model Updating (importance: 10, dev_relevance: 10)
DEX: Digit-Level Early Exit for Energy-Efficient MSDF Neural Network Inference (importance: 15, dev_relevance: 15)
Self-improving data synthesis can scale embedding-model adaptation across domains and tasks (importance: 25, dev_relevance: 40)
I3L: Iterative Interactive Imitation Learning for Whole-Body Mobile Manipulation (importance: 20, dev_relevance: 15)
PriorityFEC: Deadline-Aware Streaming FEC for KV-Cache Handoff in Edge Multi-Agent Systems (importance: 20, dev_relevance: 45)
DialFork: Going Beyond Factuality to Evaluate Procedural Hallucination in Conversational Agents (importance: 25, dev_relevance: 30)
PACE: Pareto-Adaptive Compression for Efficient Native MLLMs (importance: 20, dev_relevance: 30)
A Unified Attack Framework against Statistical Generative Watermarks for Large Language Models (importance: 30, dev_relevance: 20)
Superintelligent Software Engineering: Governance That Scales with Capability (importance: 35, dev_relevance: 70)
SOPO: Socratic Guided Policy Optimization for Span-Level Hallucination Detection (importance: 20, dev_relevance: 25)
Can Agentic Code Review Catch Vulnerabilities Before They Are Merged? (importance: 40, dev_relevance: 70)
Hierarchical Cooperative GFlowNets (importance: 15, dev_relevance: 10)
Adapting LLM-SC to Living Sacred Heritage: Domain-Specific Extraction, Semantic Networks and Human Evaluation on Pashupatinath Temple Reviews (importance: 15, dev_relevance: 20)
I-HITL: Inalienable Human-in-the-Loop (importance: 35, dev_relevance: 45)
When Is a Relation Worth Trusting? Manipulation Costs, Exposure, and the Trust Trap in Multi-Relational Graph Learning (importance: 15, dev_relevance: 15)
DeFT-RAG: Decomposing Multi-Hop Questions into Self-Contained Sub-Questions with Reusable Answers (importance: 25, dev_relevance: 40)
Making Sense of Audience Change: Causal Attribution and Boundary Management on Xiaohongshu (importance: 10, dev_relevance: 5)
Turing LLM Post-Training: Calibrated Distillation and Efficient Dynamic-MoE Execution (importance: 35, dev_relevance: 35)
NECA: Neighbor Event Curve Augmentation for Multi-Horizon ICU Deterioration Prediction (importance: 15, dev_relevance: 10)
Automatic Construction of Strong Simple Baselines (importance: 25, dev_relevance: 35)
Don't Just Impute, Detect: Task-Aware Diffusion Pipelines for Edge Anomaly Detection (importance: 15, dev_relevance: 15)
Unsupervised Feature Selection Using Local Intrinsic Dimensionality -- Extended Version (importance: 10, dev_relevance: 10)
Certified Behavioral Retention for Continual Adaptation with Held-Out Anchors (importance: 20, dev_relevance: 25)
Frozen in Time: How a Physics Foundation Model Forgets Under Sim-to-Real Adaptation (importance: 15, dev_relevance: 15)
FARO: Function-Space Adaptive Return-to-Risk Optimization via Cross-Block Equal-Risk-Contribution Budget Allocation (importance: 25, dev_relevance: 40)
Near-Optimal First-Order Regret for Bandits with Feedback Graphs (importance: 10, dev_relevance: 10)
TensorSSM: Memorizing Spatial Position in Dynamic State Space for Tensor Core Generation (importance: 15, dev_relevance: 35)
No Channels Left Behind: Understanding Diffusion Acceleration through Residual Gates (importance: 15, dev_relevance: 20)
Fair Recommender Systems for Intersecting Groups (importance: 10, dev_relevance: 15)
Monte Carlo Random Walk Kernel (importance: 10, dev_relevance: 25)
RC Prediction: A Python Package for Parameter-Aware Reservoir Computing (importance: 15, dev_relevance: 60)
When Can Unlabeled Data Pools Replace Membership Queries? (importance: 5, dev_relevance: 5)
MOSAIC: Addressing Disproportionate Task-Modality Dependencies via Modality-Partitioned Gradient Routing (importance: 15, dev_relevance: 15)
Density-Guided Bootstrap Boundaries for Latent World-Model Reinforcement Learning (importance: 15, dev_relevance: 15)
Omniprediction for All Mixable Proper Losses and Beyond (importance: 8, dev_relevance: 5)
When can Optimally Trained Transformers Achieve Near-Lossless Token Prediction for non-Markovian Data? (importance: 15, dev_relevance: 10)
Before We Adapt: The Zero-Shot Generalization Bottleneck in Continual World Models (importance: 15, dev_relevance: 10)
Localizing predictive concepts encoded in neural representations via hierarchical testing (importance: 10, dev_relevance: 15)
Communication-Efficient Private Federated Reinforcement Learning with Post-Quantum Authentication for Vehicular Networks (importance: 10, dev_relevance: 10)
Nonlinear Maps Between Irreducible Representations in Graph Neural Networks (importance: 10, dev_relevance: 10)
TalentRL: Strategy-Guided Reinforcement Learning for Hierarchical Skill Routing (importance: 20, dev_relevance: 40)
CAP: Learning to Optimize from Sparse Observations of Related Tasks (importance: 10, dev_relevance: 15)
General Frequency Principle for Linear Approximation Classes (importance: 8, dev_relevance: 5)
Role-Aware Server Optimization for Multimodal Federated Learning with Missing Modalities (importance: 10, dev_relevance: 10)
Beyond Scarcity: When Encoder Adaptation Is Worthwhile (importance: 10, dev_relevance: 10)
A Positional Realizability Theory of Length Generalization in Single-Head Attention (importance: 15, dev_relevance: 15)
Corroborative Learning: Mutual Enhancement through Task-Response Reconstruction (importance: 15, dev_relevance: 30)
Generative SDEs with Trainable Diffusion and Density-Based Losses (importance: 15, dev_relevance: 25)
A Unified Framework for Probability Estimation of Data-Dependent Events (importance: 5, dev_relevance: 20)
SPT-Flow: One Integration, Many Futures in Continuous-Time World Models (importance: 20, dev_relevance: 25)
rl_rob: A transparent, from-scratch Python library for teaching reinforcement learning from tabular methods to deep policy gradients (importance: 15, dev_relevance: 60)
StrictPRMBench: A Replayable Evaluation Benchmark for Auditing Fixed-Score PRM Aggregation (importance: 20, dev_relevance: 35)
Quantile Propagation Planning for Budgeted Tail-Risk Evaluation (importance: 10, dev_relevance: 20)
Generated storm lifecycles fill the rare-event tail of short rainfall records (importance: 15, dev_relevance: 15)
What We Keep of What We Forget: Replacement Operators and Routing Semantics in Training-Free MoE Expert Pruning (importance: 25, dev_relevance: 45)
Anomalies as Residuals: Cauchy Displacement Networks (importance: 12, dev_relevance: 35)
Certifying Retraining Regret through Residual Structure (importance: 8, dev_relevance: 25)
A Unified XAI-NSAI Analytical Framework for Trustworthy IoT Anomaly Detection (importance: 12, dev_relevance: 35)
When Everyone Can Build a Model: Mechanistic Inference in the Age of AI (importance: 20, dev_relevance: 40)
Improving the Optimization Landscape of Matrix Completion with $\epsilon$-close Surrogates (importance: 10, dev_relevance: 20)
NormDetect: Anchored Calibration for Low-False-Positive Detection of Machine-Generated Web Text (importance: 25, dev_relevance: 40)
RetryRoute: Retry-Aware Model Routing for Agentic Image Editing Services (importance: 30, dev_relevance: 65)
When Links Mislead: Label-Free Structural Polarity for Zero-Shot Classification of Web Pages (importance: 10, dev_relevance: 25)
Spectral Haar Binarizer: Compact Boolean Contrast Representations for Hyperspectral Classification (importance: 8, dev_relevance: 20)
PCD: Predictive Coding Dynamics for Irregularly Sampled Time Series (importance: 12, dev_relevance: 30)
Infinite ResNets are Gaussian Mean ODEs (importance: 12, dev_relevance: 15)
Enhancing Sentiment Classification on Small Datasets through Data Augmentation and Transfer Learning (importance: 15, dev_relevance: 35)
FOCUS: First-Order Control Using Optimal Transport Supervision for Imitation from Observations (importance: 10, dev_relevance: 25)
A Constructive Orbit-Based Theory of Learnability (importance: 5, dev_relevance: 10)
On the Generation Order in GFlowNets (importance: 10, dev_relevance: 20)
Does Reasoning Make Language Models Agree? (importance: 25, dev_relevance: 25)
From WSIs to Bulk, Spatial and Single-Cell Transcriptomics via Multi-Scale Graph Neural Networks (importance: 12, dev_relevance: 15)
From fleet trajectories to single-cell promises: hierarchically weighted conformal prediction for battery SOH monitoring (importance: 8, dev_relevance: 25)
Hierarchical Residuals: A Minimalist Decoupling of Residual Width from Compute Width (importance: 25, dev_relevance: 40)
UpLoop: Training Looped Transformers via Upcycling at No Extra Pretraining Compute (importance: 22, dev_relevance: 35)
Leakage-Controlled Structure-Preserving Augmentation and Template-Aware Evaluation for Bangla Symptom-Based Disease Prediction (importance: 5, dev_relevance: 15)
Locality-Controlled Out-of-Distribution Guidance for Robust Regulatory DNA Sequence Design (importance: 10, dev_relevance: 15)
Task Vectors Can Be Linearly Decomposed into Interpretable and Controllable Components (importance: 22, dev_relevance: 35)
Do LLMs Know When They Are Wrong About Molecules? In-Context Property Prediction with Self-Reported Uncertainty (importance: 20, dev_relevance: 25)
On-device Detection of Wearer-induced NLOS on Chest-worn UWB Tags: Machine Learning, Link Indicators, or Both? (importance: 8, dev_relevance: 40)
Towards Out-of-Distribution Uncertainty Quantification for Machine-Learned Interatomic Potentials via Pre-Training (importance: 12, dev_relevance: 20)
Exploring Bad Minima in Transfer Learning (importance: 10, dev_relevance: 15)
Continual Reinforcement Learning Is an Integration Problem (importance: 20, dev_relevance: 25)
DIFFNAS: Predictor-Free Neural Architecture Search via Diffusion Model and Zero-Cost Evaluation (importance: 15, dev_relevance: 35)
ElTab: Elastically Looped Tabular Transformers (importance: 15, dev_relevance: 30)
Trainable Vector Quantization for Large Language Models (importance: 30, dev_relevance: 50)
On k-nearest-neighbor regression in metric spaces (importance: 5, dev_relevance: 15)
Binary Sparse Codes: Where are we on Energy Efficient Representations? (importance: 10, dev_relevance: 15)
Robust Learning of Multi-Index Models under Logconcave Distributions via Weak Partial Learners (importance: 5, dev_relevance: 10)
New Lower Bounds for Private PAC Learning (importance: 5, dev_relevance: 10)
Less Loss, Fixed Weights: Logit Corrections for Forgetting (importance: 25, dev_relevance: 40)
When Does Jacobian Fine-Tuning Stabilize Neural PDE Emulators? (importance: 10, dev_relevance: 15)
GhostPrefix: Synchronization-Free Universal Prefix Attacks for ASR Manipulation (importance: 20, dev_relevance: 25)
CausalXRL: Explainable Reinforcement Learning through Causal Graph Reasoning (importance: 12, dev_relevance: 25)
A richness-sensitive Acoustic Biodiversity Index from species-level detections: classifier benchmarking, false-positive control and validation against standard soundscape indices in an African savanna and an urban zoo (importance: 3, dev_relevance: 10)
Differentially Private Synthetic Data from Sliced Chebyshev Moments (importance: 8, dev_relevance: 25)
Cost-Aware Stopping for Freeze-Thaw Bayesian Optimization: When Does One-Step Lookahead Suffice? (importance: 8, dev_relevance: 25)
Towards Understanding the Impact of Plasticity Loss on Reinforcement Learning in Stochastic Environments (importance: 15, dev_relevance: 25)
Hyperbolic Inverse Reinforcement Learning (importance: 10, dev_relevance: 15)
Fast Privacy Loss Accounting for Random Allocation via Multiscale FFT (importance: 10, dev_relevance: 40)
Gradient Descent on Two ReLU Neurons: Global Landscape and Bifurcation Dynamics (importance: 10, dev_relevance: 15)
Physiological Trajectories and Observation Patterns in ICU Mortality Prediction: Transportability Under Changing Measurement Conditions (importance: 10, dev_relevance: 15)
Instance-dependent sequential limits as an exact information ledger (importance: 5, dev_relevance: 10)
Counting on Dummies: Privacy Amplification with Observable Counts and Truncated Batches (importance: 8, dev_relevance: 30)
Neural Networks Based on Power Method and Inverse Power Method for Solving Linear Eigenvalue Problems (importance: 8, dev_relevance: 10)
A small HER2 reader for routine H&E slides: real-world deployment (importance: 25, dev_relevance: 20)
Intervention-Conditioned Counterfactual Human Digital Twin Generator (importance: 12, dev_relevance: 15)
VitisGP: Multi-Task Benchmark for Genotype to Phenotype Grapevine Prediction (importance: 5, dev_relevance: 20)
Dual-Corrected Stochastic Block Model ($\text{DC}^2\text{SBM}$): A Generalized Structure Detection Method for Attributed Networks (importance: 5, dev_relevance: 15)
CalibCause: A Quality-Gated, Interpretable and Calibrated Deep Learning Framework for Diabetic Retinopathy Grading (importance: 12, dev_relevance: 20)
FGSM Attack Transferability from Simulation to Quantum Hardware in Network Intrusion Detection (importance: 12, dev_relevance: 25)
The Geometry of Multitask Learning: Sharp Rates for Shared Subspace Estimation (importance: 5, dev_relevance: 10)
INLA for Adaptive UAV Sampling of Radio Environment Maps in IoT Networks: Prior Width Governs Accuracy and Calibration (importance: 8, dev_relevance: 25)
Li$^{2}$SAO: Lifting Light Singl-Rollout Asynchronous RL Optimization via Group-Relative Actor-Critic Bootstrapping (importance: 25, dev_relevance: 50)
Model-Agnostic Search for New Physics in the Presence of Imperfect Background by Smearing the Data (importance: 0, dev_relevance: 10)
Apriel Voice 27B Preview: An Open Model for Speech, Reasoning, and Tool Use (importance: 35, dev_relevance: 55)
Mage-VL-Cowork: A Full Training Recipe to Learn Hybrid Computer Use From Evolved Harnesses (importance: 35, dev_relevance: 60)
Combinatorial Residual Network: a decay-chain-aware classifier without explicit jet assignment (importance: 2, dev_relevance: 10)
Predictive Maintenance at the Edge: An Embedded Anomaly Detection Pipeline Trained with Quantum Optimization (importance: 8, dev_relevance: 35)
Robust-by-Design Distributional Learning from Contaminated Samples (importance: 8, dev_relevance: 25)
AquaProbe: Efficient Adaptation of Frozen Vision Models for Underwater Images (importance: 12, dev_relevance: 25)
Learning Stabilizing Latent Feedback Controllers via Multi-Step Backward Losses (importance: 6, dev_relevance: 15)
Fewer Heads, Fewer Memories: Pruning Redundant Attention Heads Suppresses Verbatim Memorisation (importance: 22, dev_relevance: 30)
Meshless Latent Dynamics Models for Model Order Reduction of time-dependent Partial Differential Equations (importance: 6, dev_relevance: 15)
Single-Annotation, Multi-Hypothesis Segmentation with Semantic-Spatial Residual Banks (importance: 10, dev_relevance: 20)
D2ML: Models for Direct-to-Biology Small Molecule Screening (importance: 12, dev_relevance: 15)
Stochastic and Noisy Oracles in Smooth Continuous Optimization (importance: 5, dev_relevance: 15)
Adaptive Uncertainty Quantification on Simplicial Complexes with Conformalized Simplicial Neural Networks (importance: 8, dev_relevance: 15)
Mechanistic Interpretability for End-to-End Self-Driving (importance: 15, dev_relevance: 20)
When AI Investigators Cannot Trust Their Tools: Conflict-Aware Evidence Verification for Cyber Threat Investigation (importance: 20, dev_relevance: 35)
Shifting Focus by Reliability: Geometric Fusion Modulation for Robust RGB-T Detection (importance: 8, dev_relevance: 25)
Coarse-Graining Optimal Transport for Rollout-Free Post-Training of Flow Matching (importance: 12, dev_relevance: 25)
AuditFL: A Contract-Backed Clearing Layer for Open Federated Learning (importance: 8, dev_relevance: 35)
Estimation of Quantum Umlaut Information (importance: 0, dev_relevance: 10)
Riemannian Optimization on Unknown Manifolds via Stochastic Proximal Refinement (importance: 5, dev_relevance: 10)
DUGO: Dual-Scale Gaussian Memory for Online Temporal Action Localization in Egocentric Procedural Videos (importance: 10, dev_relevance: 20)
Conditional Neural SDEs for Financial Time-Series Generation (importance: 8, dev_relevance: 15)
SACQ: Structure-Aware Codebook Quantization for Ultra-Low-Bit Vision-Language-Action Models (importance: 15, dev_relevance: 35)
A prospective randomized study of patient-centered AI for clinical visit preparation (importance: 20, dev_relevance: 10)
KV-Cache Patching: Tracing How and Where Fine-Tuned Behaviors Arise (importance: 15, dev_relevance: 30)
DAPS: Dependency-Aware Premise Selection for LLM Theorem Proving (importance: 15, dev_relevance: 35)
DeepWeave: Weaving Neural and Symbolic Evidence for ML-Based Network Intrusion Detection System (importance: 8, dev_relevance: 25)
BYOD: Build Your Own DLM - Efficient Autoregressive-to-Diffusion Language Model Conversion (importance: 22, dev_relevance: 40)
Reliability-Weighted Direct Preference Optimization: Penalized Weights for Robust Preference Learning (importance: 18, dev_relevance: 40)
PDE-Residual-Informed Kriging (importance: 5, dev_relevance: 15)
Underdamped Stochastic Optimal Transport with Landauer boundary conditions: machine learning and analytical solution (importance: 3, dev_relevance: 5)
What Carries Architectural Signal? A Code-to-Architecture Mapping Study (importance: 35, dev_relevance: 70)
Co$^2$ODE: Conditionally Coarse-Grained Model Discovery for Rewiring in Biochemical Networks (importance: 15, dev_relevance: 5)
Enhancing Medical Lesion Segmentation via Several Simple Modules (importance: 25, dev_relevance: 10)
$\nabla$Mimic: Ultra-Efficient Humanoid Control with Differentiable Simulation (importance: 30, dev_relevance: 15)
Leveraging Quantum Measurement Randomness for Shot-Efficient Diffusion Models (importance: 20, dev_relevance: 5)
From Training Exposure to Membership Evidence: A Controlled Study of OLMo 2 (importance: 35, dev_relevance: 30)
Peeling the Matryoshka: Extracting Ontological Structure from Retrieval Embeddings (importance: 30, dev_relevance: 35)
What Standard Evaluations Hide About Detectors of Hijacked LLM Agents: Rare Attacks, Unseen Tools and Benchmark Transfer (importance: 40, dev_relevance: 50)
Overcoming the Representational Limitations of Amplitude Encoding for Quantum Neural Networks (importance: 20, dev_relevance: 5)
Text as an Index for Zero-Shot Cross-Domain Knowledge Tracing (importance: 20, dev_relevance: 15)
Learning Nonlocal Interactions in Large-Scale Heterogeneous Metamaterials through Scientific Machine Learning (importance: 15, dev_relevance: 10)
On-Chain Behavioral Signatures for Classifying Centralized Exchange Addresses on Ethereum Using Proof-of-Reserves (importance: 15, dev_relevance: 15)
Statistical Foundations of Reinforcement Learning: A Non-Asymptotic Perspective (importance: 25, dev_relevance: 15)
Variance Reduction for Decentralized Stochastic Optimization with Multiple Local Updates (importance: 15, dev_relevance: 10)
Shock Loading Accesses Reaction Pathways beyond Accelerated Thermolysis (importance: 20, dev_relevance: 5)
WASABI: Wasserstein Sampling via Bridge Interpolants (importance: 20, dev_relevance: 5)
When Is a Game a Markov Decision Process? Correspondences Between Game Theory and Reinforcement Learning (importance: 25, dev_relevance: 20)
A Neural Network Approach towards Real-Time Knot Point Density Estimates for the Nonlinear Quadrotor (importance: 15, dev_relevance: 25)
Empirical Evaluation of Quantum Kernel Methods for Classification: A Comparative Study with Classical Kernels (importance: 20, dev_relevance: 10)
Exact Theory and Optimal Stepsizes for Stochastic Gradient Descent in Smooth Convex Minimization (importance: 30, dev_relevance: 25)
RAMA: Retrieval-Augmented Multi-Agent Routing with Episodic Memory for LEO Satellite Networks (importance: 15, dev_relevance: 15)
Hard Constraint Projection in a Physics Informed Neural Network (Burgers) (importance: 15, dev_relevance: 10)
Regret-Optimal Wasserstein-Robust Linear Regression (importance: 15, dev_relevance: 10)
SATFRONT: Detecting Atmospheric Fronts with Satellite Data using Machine Learning (importance: 20, dev_relevance: 10)
A Better Critic Matters for Distribution Matching Distillation (importance: 30, dev_relevance: 20)
Latent Reasoning in Continuous Space for Unified Multimodal Models (importance: 35, dev_relevance: 15)
Bilevel ZOFO: Efficient LLM Fine-Tuning and Meta-Training (importance: 30, dev_relevance: 50)
Gaussian Process Upper Confidence Bound Achieves Nearly Optimal Regret in Noise-Free Gaussian Process Bandits (importance: 15, dev_relevance: 10)
Slim attention: cut your context memory in half without loss -- K-cache is all you need for MHA (importance: 35, dev_relevance: 55)
Non-Asymptotic Stability and Consistency Guarantees for Physics-Informed Neural Networks via Coercive Operator Analysis (importance: 15, dev_relevance: 10)
Who Gets Watched? Allocating Scarce Escalation Resources on Pediatric Wards with Neural Index Policies (importance: 15, dev_relevance: 5)
Geometry as a Missing Axis of Representation Quality: The Variational Geometric Information Bottleneck under Data Scarcity (importance: 15, dev_relevance: 10)
Learning Robust Social Strategies with Large Language Models (importance: 25, dev_relevance: 15)
Temporal speckle enables probabilistic weights for uncertainty-aware photonic AI (importance: 25, dev_relevance: 5)
Natural Language Actor-Critic: Scalable Off-Policy Learning in Language Space (importance: 35, dev_relevance: 40)
A Learning Stability Profile for Finite-Dimensional Learning Dynamics (importance: 15, dev_relevance: 15)
Fusion or Confusion? Multimodal Complexity Is Not All You Need (importance: 25, dev_relevance: 15)
Your Group-Relative Advantage Is Biased (importance: 35, dev_relevance: 35)
Sample-wise Constrained Learning via a Sequential Penalty Approach with Applications in Image Processing (importance: 15, dev_relevance: 10)
Federated learning for unpaired multimodal data (importance: 25, dev_relevance: 15)
SmartMeterFM: Flow Matching for Smart Meter Data Generation, Imputation, and Super-Resolution Without Retraining (importance: 15, dev_relevance: 10)
Root Cause Analysis of Measurement and Mechanistic Anomalies (importance: 15, dev_relevance: 15)
A Unified Matrix-Spectral Framework for Stability and Interpretability in Deep Learning (importance: 15, dev_relevance: 15)
A Function-Space Stability Boundary for Generalization in Interpolating Learning Systems (importance: 15, dev_relevance: 10)
When Less is More: The LLM Scaling Paradox in Context Compression (importance: 35, dev_relevance: 40)
Discrete Diffusion Models Exploit Asymmetry to Solve Lookahead Planning Tasks (importance: 25, dev_relevance: 15)
Spectral Condition for $\mu$P under Width-Depth Scaling (importance: 35, dev_relevance: 40)
AIMER: Calibration-Free Task-Agnostic MoE Expert Pruning (importance: 30, dev_relevance: 35)
A Theory-guided Weighted $L^2$ Loss for solving the BGK model via Physics-informed neural networks (importance: 10, dev_relevance: 5)
When Does Context Help? A Systematic Study of Target-Conditional Molecular Property Prediction (importance: 15, dev_relevance: 5)
Towards Universal Tabular Embeddings: A Benchmark Across Data Tasks (importance: 25, dev_relevance: 25)
ORACLE: Anticipating Scams from Partial Trajectories in Streaming App Usage (importance: 20, dev_relevance: 10)
Multiscale Supervised Unbalanced Optimal Transport Flow Matching (importance: 15, dev_relevance: 10)
Step-wise Rubric Rewards for LLM Reasoning (importance: 30, dev_relevance: 30)
Deep Neural Sheaf Diffusion (importance: 15, dev_relevance: 10)
Cyclical Entropy Eruption: Entropy Dynamics in Agent Reinforcement Learning (importance: 30, dev_relevance: 40)
Conveyance: A Versatile Framework for Learning in Structured Class Spaces (importance: 15, dev_relevance: 15)
Query Entity-Oriented Graph Neural Networks for Knowledge Graph Completion (importance: 15, dev_relevance: 10)
How to Score Experts for One-Shot MoE Expert Pruning: A Systematic Formulation and Selection Principle (importance: 25, dev_relevance: 35)
AME: A Multi-Type Contributor Attribution Framework in Generative AI Markets (importance: 20, dev_relevance: 15)
MIRAGE: Auditing Anti-Muslim Bias in Frontier LLMs Across Reasoning, Agentic, and Time-Coupled Conditions (importance: 25, dev_relevance: 15)
On the Design Space of Discrete Diffusion Online Adaptation for Molecular Optimization (importance: 15, dev_relevance: 5)
Windowed-MTP: Removing the Full-Context Draft-KV Tax at Million-Token Context (importance: 35, dev_relevance: 60)
Bayesian Complete-Pooling in Cross-Subject Classification for Motor Imagery Electroencephalogram (importance: 15, dev_relevance: 5)
MS-GPT: Rethinking MS/MS De Novo Structure Elucidation as Spectrum-Induced Posterior Querying of a Molecule-Language Model (importance: 15, dev_relevance: 5)
Evaluating Rollout-Verified Reinforcement Fine-Tuning of a Reasoning Model for Multi-Zone Variable-Air-Volume Control (importance: 25, dev_relevance: 20)
Inadvertent Context Leakage in Language Models (importance: 35, dev_relevance: 25)
Predicting Dose-Response Success from Primary Screening Data (importance: 10, dev_relevance: 5)
Toward Optimal Switching Regret for Multi-Armed Bandits with Oblivious Adversary (importance: 10, dev_relevance: 5)
Non-Commutative Lie-Algebraic State-Space Models: Finite-Division Variety Laws, Semidirect Cascades, and Gradient Saddle Escape (importance: 20, dev_relevance: 20)
SkillEvoLean: Mutation-enhanced skill evolution for Lean provers (importance: 25, dev_relevance: 35)
Sharp Non-Asymptotic Analysis of the Penalized Challenger in $\beta$-EB-TCI for Bernoulli Bandits (importance: 10, dev_relevance: 5)
Are We Really Benchmarking Forecasting Models? The Impact of Preprocessing on Time Series Performance (importance: 20, dev_relevance: 15)
4-Tensor Attention Model for Semantic Physical Reality (importance: 20, dev_relevance: 10)
TILE: Bridging LLMs and Table Foundation Models for Tabular Learning with Strings (importance: 20, dev_relevance: 20)
CodeLift: Your Test Suite Is Secretly a Dense SWE Curriculum (importance: 35, dev_relevance: 55)
Transport What Shares the Conduction Pathway: Normal-Anchored SPD Geometric Transport for Source-Free Patient Adaptation of Holter and Patch ECG (importance: 15, dev_relevance: 5)
Estimating Performance for Safe Deployment of Machine Learning Models (importance: 20, dev_relevance: 20)
When Does Dynamics Randomization Test Context Inference? (importance: 15, dev_relevance: 10)
Satellite-derived bathymetry of Arabian Gulf ports under leakage-free validation: a multi-site benchmark of empirical, machine-learning and deep-learning models against 285,000 multibeam soundings, with an outlook for UAE waters (importance: 10, dev_relevance: 5)
Cumulative Marginal Matching for One-Step Diffusion (importance: 25, dev_relevance: 10)
h1: A Taste of the Bitter Lesson in Automated Theorem Proving (importance: 30, dev_relevance: 35)
Normative Deviation Modeling for EEG-based Depression Detection (importance: 10, dev_relevance: 5)
Heads of Security: Learned Attention Gates Improve Prompt Injection Robustness (importance: 35, dev_relevance: 45)
TaxoMFM: A Taxonomy-Aware Foundation Model for the Human Gut Microbiome (importance: 10, dev_relevance: 5)
Discovering contemporaneous and lagged causal relations in autocorrelated nonlinear time series datasets (importance: 10, dev_relevance: 10)
Transport-Free Lower Bounds for Ollivier--Ricci Curvature in Higher-Order Networks: Scope, Tightness, and Computational Applications (importance: 10, dev_relevance: 5)
Learning Maximal Regions of Attraction on Compact Sets using Taylor Lyapunov Networks (importance: 15, dev_relevance: 10)
Lattice-Based Pruning of Recurrent Neural Networks via Poset Modeling: A Corrected Theory and a Matched-Sparsity Evaluation (importance: 10, dev_relevance: 15)
Critically-Damped Higher-Order Langevin Dynamics for Generative Modeling (importance: 20, dev_relevance: 10)
Distributed Information Bottleneck Theory for Multi-Modal Task-Aware Semantic Communication (importance: 10, dev_relevance: 5)
DNNs, Dataset Statistics, and Correlation Functions (importance: 15, dev_relevance: 10)
Repair Brain Damage: Real-Numbered Error Correction Code for Neural Network (importance: 15, dev_relevance: 15)
Parabolic Positional Encoding: A Vision-Centric, Principled, Extrapolatable, and General Approach (importance: 15, dev_relevance: 25)
Expected Harm: Rethinking Safety Evaluation of (Mis)Aligned LLMs (importance: 20, dev_relevance: 15)
Cross-Fitted Clipped Covariance Estimation with a Data-Driven Tail-Energy Criterion (importance: 10, dev_relevance: 10)
Co-optimization for Adaptive Conformal Prediction (importance: 10, dev_relevance: 10)
FlowTouch: View-Invariant Visuo-Tactile Prediction (importance: 15, dev_relevance: 10)
A Controlled Comparison of Deep Learning Architectures for Multi-Horizon Financial Forecasting: Evidence from 918 Experiments (importance: 15, dev_relevance: 15)
Understanding the Performance Plateau in Text-to-Video Retrieval: A Comprehensive Empirical and Linguistic Analysis (importance: 15, dev_relevance: 10)
A Cross-Layered Multi-Drone Coordination for Medical Supply Delivery during Disaster Response Management (importance: 15, dev_relevance: 10)
Quantifying Concentration Phenomena of Mean-Field Transformers in the Low-Temperature Regime (importance: 35, dev_relevance: 25)
Events as Triggers for Behavioral Diversity in Multi-Agent Reinforcement Learning (importance: 25, dev_relevance: 10)
Optimal Design for Multinomial Logit Model with Applications to Best Assortment Identification (importance: 20, dev_relevance: 10)
Geometry of Relaxed Fair Regression: A Unified Framework for Aware and Unaware Settings (importance: 20, dev_relevance: 8)
Accelerating Min-Max Optimization via Power-Law Stepsizes (importance: 20, dev_relevance: 10)
No-Harm Physics-Informed Inverse Learning with Residual-Calibrated Uncertainty (importance: 20, dev_relevance: 10)
Soft Mixture-of-Recursions: Going Deeper with Recursive Vision Transformers (importance: 25, dev_relevance: 10)
Fields of the Planet: Field Boundary Mapping Beyond 10m (importance: 20, dev_relevance: 10)
How Much Reconstruction Does Quantum Machine Learning Need? Late Fusion of Uncoupled Quantum Subcircuits (importance: 15, dev_relevance: 15)
You Only Pass Once: Answering and Abstaining Together in a Single Forward Pass of a Frozen Language Model (importance: 35, dev_relevance: 25)
Minimax Alternating Regret for the Experts Problem and Online Convex Optimization (importance: 15, dev_relevance: 5)
A Nearly Quadratic Lower Bound for Linear Optimization over Convex Bodies in the Membership Oracle Model (importance: 0, dev_relevance: 5)
FLAT: Smoothing the Rugged Landscape for Learnable, Sample-Efficient Traffic Calibration (importance: 15, dev_relevance: 8)
Instrument-Anchored LLM Judges Transfer Expert Counseling-Quality Ratings Across Three German Corpora (importance: 20, dev_relevance: 15)
Diversity-Aware Ensemble of Quantum-Classical Models via Combinatorial Fusion (importance: 10, dev_relevance: 5)
Gluing Local Contexts into Global Meaning: A Sheaf-Theoretic Decomposition of Transformer Representations (importance: 25, dev_relevance: 15)
Accelerated Dataflow: scaling inference in streaming data analytics with efficiency (importance: 25, dev_relevance: 45)
Recursive Code Evolution for Task-Oriented Semantic Communications (importance: 15, dev_relevance: 25)
From GPU to TPU with Coding Agents: What Works, What Fails, and What It Costs? (importance: 42, dev_relevance: 70)
An Agentic AI Workflow for Liver Lesion Segmentation and Structured Radiology Reporting (importance: 20, dev_relevance: 10)
PDRVLA: A Vision-Language-Action Model with Physical Dynamics Representations (importance: 25, dev_relevance: 5)
Towards Reliable, Scalable, and Value-Aligned Generative Recommendation for Large-Scale Advertising (importance: 25, dev_relevance: 10)
GROME: GROunding Factorized MEmory into Video World Models (importance: 25, dev_relevance: 5)
An Empirical Study of Gas Optimization Practices in Solidity Smart Contracts (importance: 0, dev_relevance: 55)
AIFaultBench: A Reproducible Benchmark of Real-World Faults in AI Software Systems (importance: 40, dev_relevance: 80)
RepGen: Reproducing Deep Learning Bugs through Context-Aware Code Generation and Verification (importance: 35, dev_relevance: 80)
Cognitive Inclusivity in AI Coding Assistants: Evidence from GitHub Copilot (importance: 35, dev_relevance: 70)
Yesterday's Code, Today's Dependencies: How Dependency Selection Shapes PHP Migration Benchmarks (importance: 12, dev_relevance: 65)
Architectural Debt Enforcement under Incomplete Observation: A Controlled Study of Ratchets and Evidence Qualification (importance: 0, dev_relevance: 60)
What Does a Repair Agent Gain from Remembering Its Failure? An Exploratory Study on Real Python Bugs (importance: 40, dev_relevance: 80)
Post-SQLFix: Property-Guided Repair of LLM-Generated SQL with Runtime Verification (importance: 25, dev_relevance: 65)
SCRBench: Beyond Equivalence Checking for Semantic Change Reasoning in Code (importance: 25, dev_relevance: 60)
Prompt Less, Infer More: Static Analysis-Guided LLM Type Annotation for Python (importance: 25, dev_relevance: 70)
A Failure Mode Taxonomy and Severity Model for Ethereum Smart Contracts (importance: 0, dev_relevance: 50)
Reparo: Synthesizing and Validating Access Control Policies (importance: 25, dev_relevance: 60)
Stuck on Red? Understanding GitHub Actions Failing Streaks (importance: 0, dev_relevance: 70)
No More Import Errors: Constrained Decoding for Guaranteed Valid Imports in LLM-Generated Tests (importance: 30, dev_relevance: 75)
Dittany: Contribution-Guided Repair Framework for MAS Applications (importance: 30, dev_relevance: 65)
Context-Specific Instruction: A Longitudinal Study on Debugging Skill Acquisition and Retention for Novice Programmers (importance: 0, dev_relevance: 40)
Human-AI Synergy in Agentic Code Review (importance: 40, dev_relevance: 75)
Measuring Delivery Consistency in Practice: A DORA Extension from a Multi-Platform Release Setting (importance: 0, dev_relevance: 65)
STORY2DIFF: A Dataset Linking Real-World User Stories to Code Changes (importance: 20, dev_relevance: 60)
PR4Code: A Pull Requests Dataset for AI Code Generation (importance: 25, dev_relevance: 65)
Jailbreaking my car's navigation system just so I can change the time (importance: 0, dev_relevance: 20)
An outsider's perspective on Scion and Mononoke (importance: 0, dev_relevance: 60)
What are you doing this week? (importance: 0, dev_relevance: 5)
Postgres 19 is late, and it's AI's fault (importance: 35, dev_relevance: 70)
My reflex of assuming our servers need ever-growing amounts of RAM (importance: 0, dev_relevance: 25)
Work on GEFS (the Good Enough File System) for OpenBSD progresses (importance: 0, dev_relevance: 40)
Polyend releases the Tracker firmware source code (importance: 0, dev_relevance: 35)
Pratt parsing is a black box (importance: 0, dev_relevance: 55)
SuperVote: 15 year retrospective on an app for sharing political endorsements (importance: 0, dev_relevance: 20)
A400000 Winners (importance: 0, dev_relevance: 5)
the lost malloc() (importance: 0, dev_relevance: 45)
Program (importance: 0, dev_relevance: 10)
London Meetup - Nov 2026 (importance: 0, dev_relevance: 10)
Recognizing Keystrokes Using WiFi Devices (importance: 0, dev_relevance: 30)
Introducing Kwi, a new KDE terminal emulator (importance: 0, dev_relevance: 50)
Building a gemini server in python (importance: 0, dev_relevance: 40)
The Beginning of BSD Unix (importance: 0, dev_relevance: 20)
gohawk | Resource-focused static analysis for Go (importance: 0, dev_relevance: 60)
Union Types and Option Types (importance: 0, dev_relevance: 55)
Securing Internet-Critical Software: How We Could Have Backdoored Bash (importance: 0, dev_relevance: 60)
The advantages of mobile Linux over AOSP (importance: 0, dev_relevance: 30)
New Release: Tor Browser 16.0 | Tor Project (importance: 0, dev_relevance: 25)
Virtual Art Museum built with Three js (importance: 0, dev_relevance: 40)
There’s a microchip in my toothbrush. But WHY? (importance: 0, dev_relevance: 10)
A Makefile Is the Best Agent Doc: 10 Targets Every Repo Needs Before an Agent Touches It (いいね相当スコア: 0) (importance: 0, dev_relevance: 80)
Google Cloud Spend Caps Reach General Availability With an API and Live Costs (いいね相当スコア: 0) (importance: 0, dev_relevance: 75)
CrewAI vs AutoGen: Enterprise Multi-Agent Coordination, Role Delegation, and Token Economics Showdown (いいね相当スコア: 0) (importance: 0, dev_relevance: 70)
AIGB: The Free Forever Telegram Bot for Indonesia & Philippines (いいね相当スコア: 0) (importance: 0, dev_relevance: 20)
Omen AI acquires Aquatic Labs in a $16M cash-and-stock deal (いいね相当スコア: 0) (importance: 0, dev_relevance: 25)
SemiAnalysis launches GDPx, an agentic benchmark across 11 professional domains (いいね相当スコア: 0) (importance: 0, dev_relevance: 35)
Gartner publishes its first Magic Quadrant for AI agent management platforms (いいね相当スコア: 0) (importance: 0, dev_relevance: 45)
Passing once is not the same as being reliable (いいね相当スコア: 0) (importance: 0, dev_relevance: 50)
Low-Cost Property Ticket Summarization — Token Reduction Strategies in Node.js (いいね相当スコア: 0) (importance: 0, dev_relevance: 60)
How to Use CuriousLM AI for Beginners (2026 Guide) (いいね相当スコア: 0) (importance: 0, dev_relevance: 10)
What Is Danmaku? How to Download Bilibili Bullet Comments and Find Any Video's Highlight Of Bilibili (いいね相当スコア: 0) (importance: 0, dev_relevance: 15)
Whoever takes your old GitHub name gets your MCP namespace (いいね相当スコア: 0) (importance: 0, dev_relevance: 85)
How to build an AI agent for production (いいね相当スコア: 0) (importance: 0, dev_relevance: 80)
Running an LLM fully offline on Android with Kotlin coroutines (いいね相当スコア: 0) (importance: 0, dev_relevance: 85)
San Francisco’s Startup Map Is Missing Something: A Map of How Innovation Actually Works (いいね相当スコア: 1) (importance: 0, dev_relevance: 10)
Building Secure LLM Applications: A Practical 14-Point Checklist Based on OWASP 2026 (いいね相当スコア: 0) (importance: 0, dev_relevance: 70)
Replacing LiteLLM in Docker (500 MB RAM) with an 11 MB Go binary: fast LLM fallbacks and OS keyring (いいね相当スコア: 0) (importance: 0, dev_relevance: 85)
Priced per TB/s, the H200 beats the H100 at 6 of 7 providers (いいね相当スコア: 0) (importance: 0, dev_relevance: 40)
Run a Capable LLM on Your Own Machine (いいね相当スコア: 0) (importance: 0, dev_relevance: 75)
Hybrid retrieval with reciprocal rank fusion: notes from a RAG prototype (いいね相当スコア: 0) (importance: 0, dev_relevance: 75)
FlatCake: Why Tiered KV Cache Storage Is Holding LLM Serving Back (いいね相当スコア: 0) (importance: 0, dev_relevance: 60)
Meituan's LongCat-2.5 Drops Into Claude Code at $0.30/$1.20. Its Setup Caps You at 200K (いいね相当スコア: 1) (importance: 0, dev_relevance: 85)
What a 32GB GPU actually lets you run locally (いいね相当スコア: 0) (importance: 0, dev_relevance: 50)
AMA Announcement: Reflection, the lab behind behind Beam (Tuesday Nov 3rd, 8am-11am PT) (importance: 25, dev_relevance: 30)
Official Statement: Qwen Models Will Stick to Open Source to the End! (importance: 40, dev_relevance: 40)
[Release] ISTA DeepSeek-V4.1-Flash-1017 GSQ-RCO Coder: a 763B model at 1 bit per parameter (importance: 40, dev_relevance: 55)
Not another tok/s post: local quants vs Claude on 9 real bugs (importance: 30, dev_relevance: 70)
Qwen 4 When ? (importance: 10, dev_relevance: 15)
Yet another reason to self host: Anthropic has (again) cut back on reverse engineering tools (importance: 0, dev_relevance: 70)
Am I the only one who frequently opens this subreddit and stumbles upon announcements about Qwen4? (importance: 0, dev_relevance: 5)
Unsloth vs GSQ-RCO - 512K context DeepSeek-V4.1 GGUF benchmarks Terminal Bench (importance: 25, dev_relevance: 60)
Proposal: A community revolution for private quality benchmarks (and why endless chasing tok/s is rotting our brains (importance: 15, dev_relevance: 35)
Running UNCENSORED DeepSeek-V4.1-Flash-1017 on a 12 Gb VRAM (importance: 30, dev_relevance: 35)
sensenova/SenseNova-Loop-9B-Preview · Hugging Face (importance: 35, dev_relevance: 30)
Fine tuned open model ftw (importance: 45, dev_relevance: 35)
Builders got MiniMax M3.1 Flash's open weights from 252.4 GB down to 93 GB in 10 days, and the lab thanked them by name (importance: 45, dev_relevance: 55)
[GLB-3D-Coder] A tiny 4B coding model that creates 3D assets for Godot and other game engines (importance: 35, dev_relevance: 55)
😃Made My Own Invoicing Dekstop App With Qwen 3.8 27B+OpenCode+LM Studio On RTX 5090..Saved Me $100 a Year I Was Paying (importance: 25, dev_relevance: 60)
We released KVARK Sparkle 1.0 – a Qwen3.8-27B-based model with up to 67% fewer thinking tokens (importance: 40, dev_relevance: 40)
4" touchscreen for inference engine monitoring (importance: 15, dev_relevance: 35)
Do we really need all these "speed" posts? (importance: 15, dev_relevance: 30)
llama + CUDA: flash_attn_ext_rows (fix unified kv for multi-seq) by am17an · Pull Request #29510 · ggml-org/llama.cpp (importance: 35, dev_relevance: 90)
An AI agent just beat NVIDIA's FlashMLA baseline by 9.8% on B200, without seeing its source code (importance: 55, dev_relevance: 75)
a 144m model makes most of my villagers' choices, gemma 4 e4b only steps in when something goes wrong (6 gb laptop gpu) (importance: 25, dev_relevance: 30)
JetBrains/Mellum2.1-12B-A2.5B-Thinking-MTP · Hugging Face (importance: 45, dev_relevance: 65)
Heretic 2.0 released: Model modification for everyone (importance: 35, dev_relevance: 50)
Weekly Hiring Thread (importance: 0, dev_relevance: 5)
Automation has taken over most of my work, so why am I still sitting in the office for 8 hours? (importance: 20, dev_relevance: 25)
How do you decide an AI agent tool is mature enough for real business work? (importance: 20, dev_relevance: 35)
AI Agents for preparing sessions (importance: 15, dev_relevance: 25)
Can a cascaded router network significantly help reduce the attack surface when running high-risk LLM agent projects? (importance: 25, dev_relevance: 60)
AI agent for Industrial automation (importance: 15, dev_relevance: 25)
I ran the same browser job 216 times with Claude Code and Codex. A script beat every agent run, and the best browser tool depended on the agent (importance: 52, dev_relevance: 80)
Google's agent team says test discrete behaviors, not just benchmark scores. How do you catch an agent getting worse after a prompt, tool or model change? (importance: 45, dev_relevance: 85)
Before automating a business workflow with AI, I'd ask one uncomfortable question. (importance: 15, dev_relevance: 20)
Anyone using AI agents for large legacy migrations (Java upgrades, framework moves) at an insurer or bank? (importance: 25, dev_relevance: 65)
Agent Graph Engineering, parts 2 to 7: splitting the services, contracts, tools, subgraphs, human in the loop (importance: 30, dev_relevance: 80)
I am building an X manager autonomous Agent, its goal is to get me more followers and start interesting discussion about my business. How do you make sure your agent is learning over time (after months)? (importance: 15, dev_relevance: 25)
Anybody has Instinct Invite ols (importance: 10, dev_relevance: 25)
The approval queue isn't the tax on your agents. It's the only part that makes them smarter. (importance: 20, dev_relevance: 30)
How to get AI presentation tools to follow negative constraints in prompts?? (importance: 10, dev_relevance: 15)
AI Agent failure research and analysis (importance: 20, dev_relevance: 60)
We found a method to make reverse engineering with agents over 700x more expensive to the attacker while adding only a small linear cost to the defender. (importance: 25, dev_relevance: 60)
They promised me Jarvis… they gave me Drew Barrymore in 50 First Dates. I will set off all your BS trip wires, but you should still try my Hermes alternative anyway. I make no money at all from it and it’s S tier good. (importance: 15, dev_relevance: 35)
centralized AI OS (importance: 10, dev_relevance: 15)
What actually broke when we put agents on back-office work (importance: 25, dev_relevance: 55)
38 Applications Filled Out - 2 Interviews - Thanks Grokbot (importance: 15, dev_relevance: 20)
How I build and deploy agents to WhatsApp in one prompt (importance: 20, dev_relevance: 40)
How are banks and fintechs actually using agents (and AI in general) today? How are they making sure its reliable? (importance: 15, dev_relevance: 25)
How To Trap Sam Altman (importance: 0, dev_relevance: 0)
Sam Altman's RAM allocation argument, applied to food (importance: 3, dev_relevance: 0)
WHAT YOU SAY!! (importance: 2, dev_relevance: 0)
Insane (importance: 0, dev_relevance: 0)
Sam Altman is right (importance: 2, dev_relevance: 0)
AI agent swarms unleashed in Minecraft to ‘build a thriving civilization’ — 50-agent teams fail after falling into paralyzing bureaucracy (importance: 40, dev_relevance: 30)
What else. (importance: 0, dev_relevance: 0)
In light of recent words of Sam Altman (importance: 2, dev_relevance: 0)
Ram Wastage (importance: 0, dev_relevance: 0)
ChatGPT stuck on “Adding pasted text...” indefinitely — anyone else experiencing this today? (importance: 15, dev_relevance: 30)
OpenAI and Anthropic are worried people will switch to local models, but they are practically asking for it at this point (importance: 25, dev_relevance: 30)
OpenAI Is Giving Up on Its Best Feature. (importance: 0, dev_relevance: 25)
In Defense is Sam. He's completely right about DRAM (importance: 3, dev_relevance: 0)
My Precious (importance: 0, dev_relevance: 0)
ChatGPT gets metered in tokens in December. The habit I'm dropping first is keeping one giant chat for each project (importance: 0, dev_relevance: 50)
Abu Dhabi agrees landmark OpenAI deals to transform government services and education (importance: 40, dev_relevance: 15)
OpenAI builds LifeVal to grade ChatGPT on childcare, budgets and paperwork (importance: 35, dev_relevance: 15)
Consent is key (importance: 2, dev_relevance: 0)
Thanks Sam (importance: 1, dev_relevance: 0)
RAM IS RAM IS RAM (importance: 0, dev_relevance: 0)
The hell happened to Astra? (importance: 25, dev_relevance: 40)
Why gtp 6 Luna? (importance: 20, dev_relevance: 35)
[D] Self-Promotion Thread (importance: 0, dev_relevance: 5)
ML PhDs: Would you rather have done your PhD before or after the rise of AI chatbots? [D] (importance: 5, dev_relevance: 5)
Small-scale check of "good views beat next-token prediction" (Balestriero & LeCun 2026): human captions only, 11M-param encoders [P] (importance: 40, dev_relevance: 25)
To all ML PhDs, how are you guys doing these days? [D] (importance: 5, dev_relevance: 5)
What did Ilya Sutskever mean when he said "The models, they just want to learn"? [D] (importance: 10, dev_relevance: 5)
Operators Day : 10 founders hiring for founding ML roles - Nov 5 [n] (importance: 5, dev_relevance: 10)
Couldn't reproduce a published paper's numbers with its own code and settings. How do I present this? [R] (importance: 25, dev_relevance: 40)
Any update with the accomodation from NeurIPS [D] (importance: 3, dev_relevance: 5)
Why do all NN optimizers have weird names? [d] (importance: 10, dev_relevance: 25)
What actually broke when I tried to fine-tune on web-scraped data and what fixed it [D] (importance: 25, dev_relevance: 70)
How to handle object deduplication across multiple camera angles using YOLO (especially for tiny 20x22px crops) across isolated user sessions? [D] (importance: 15, dev_relevance: 60)
[D] Simple Questions Thread (importance: 0, dev_relevance: 5)
Benchmarking: Qwen 27B, 35B, QFN180B; Nemotron Bolt 74B, Flint40B & Clover 35B (Q4) on M5 Max 2026 on Kapitsa with KFlash [P] (importance: 25, dev_relevance: 50)
[R] There is no free lunch in RLVR: under strategy specialization, every pass@1 gain provably costs some prompt, and the pass@k crossover has a tight bound [R] (importance: 40, dev_relevance: 30)
Are there fundamental repeating patterns in ML techniques across modalities (text, image, audio) that let us group them? [R] (importance: 10, dev_relevance: 15)
Data augmentation question. How do I characterise an original and augmented datasets where the augmentation parameters were zoom and crop? [P] (importance: 10, dev_relevance: 25)
Best option for local models [D] (importance: 10, dev_relevance: 30)
LSREP: A Longitudinal State-Replay Protocol for Evaluating Conversational Memory [R] (importance: 30, dev_relevance: 40)
Almost half of all new Steam games use AI now (importance: 30, dev_relevance: 15)
How to stop Meta, OpenAI, and Apple from training AI on your data in the UK (importance: 25, dev_relevance: 10)
OpenAI says its AI is getting better at doing things humans do, like shopping and meal planning (importance: 30, dev_relevance: 15)
Dell XPS Googlebook Review: Beautiful Hardware For A Useless Platform (importance: 10, dev_relevance: 20)
Gulf investors bet on AI, space tech (importance: 15, dev_relevance: 5)
Trump’s achilles heel is turning his own voters against him (importance: 0, dev_relevance: 0)
Unified Embodied Mind Architecture for growing a conscious mind. (importance: 2, dev_relevance: 10)
Suggestions for an Emerging Technology Theme for a College Tech Event (importance: 5, dev_relevance: 10)
let's guess what it is. here are all the clues i have for this (importance: 5, dev_relevance: 10)
Am I the only one who feels Gemini Argon is worse than even Gemini 3.1 Pro? Because I need someone to tell me I’m not losing my mind (importance: 25, dev_relevance: 30)
Michigan will pilot a prison classification system targeting 30% better violence-risk predictions starting December 1 (importance: 15, dev_relevance: 5)
Brandon sex offender had 'sexual conversations' with AI-generated children: HCSO (importance: 25, dev_relevance: 5)
Like AI, we have had access to great computing power, all types of open and closed source softwares for decades. Still most people aren't even using Excel fully. AI is capable but most people won't use it as well for most use cases (importance: 5, dev_relevance: 10)
How do you guys handle AI image/video moderation at scale? (importance: 15, dev_relevance: 60)
Feds find small businesses that use AI are more likely to hire (importance: 25, dev_relevance: 10)
‘If it wasn’t for ChatGPT my brother would still be alive’: inside OpenAI’s safety unit (importance: 35, dev_relevance: 10)
X/Twitter makes AI dev look effortless, but I keep burning tokens on unfinished projects. How are you actually shipping full apps? (importance: 15, dev_relevance: 45)
The Architectural Ultimatum (importance: 10, dev_relevance: 35)
Working on a research framework for intelligence beyond infrastructure and looking for honest feedback (importance: 3, dev_relevance: 5)
If in doubt, delete the uv.lock file (importance: 3, dev_relevance: 83)
Quoting Kyle Kingsbury (importance: 33, dev_relevance: 68)
Agent Inbox (importance: 15, dev_relevance: 35)
Help (importance: 5, dev_relevance: 10)
Xmind 2027 (importance: 35, dev_relevance: 10)
Friday Work (importance: 35, dev_relevance: 45)
When DeepSeek super agent goes quiet: diagnose one boundary at a time (importance: 15, dev_relevance: 50)
Tell HN: Mistrap Pro includes DeepSeek v4.1 Flash 1017 (importance: 25, dev_relevance: 50)
‘Coming for the King’ in AI. OpenAI, DeepSeek & Samsung. ARD #197 - AI: Reset to Zero (importance: 30, dev_relevance: 15)
DeepSeek AI USD (DEEPSEEK-USD) Seasonal returns chart - Yahoo Finance Australia (importance: 5, dev_relevance: 5)
Xiaohongshu Joins DeepSeek's $10.55 Billion Series B Round - KuCoin (importance: 75, dev_relevance: 15)
Build a RAG Pipeline With Evals: 12 Steps, 90 Min [2026] - shattered.io (importance: 30, dev_relevance: 85)
Elon Musk’s Grok AI Predicts a Huge Move for Ethereum: Can it Really Go This High by 2027? - 99Bitcoins (importance: 8, dev_relevance: 5)
Nvidia and AMD Are Backing a $6 Billion AI Startup Taking on Grok - Benzinga (importance: 72, dev_relevance: 20)
Grok vs Claude API Test: 13 Steps in 70 Min [2026] - https://tech-insider.org/ (importance: 20, dev_relevance: 70)
Elon Musk’s Grok AI Predicts Lighter Crypto to Flip Hyperliquid - Cryptonews (importance: 5, dev_relevance: 5)
Best AI Chatbot Subscription by Plan: $20 to $500 [2026] - https://tech-insider.org/ (importance: 20, dev_relevance: 15)
Elon Musk says 'unsustainably high' costs are the reason for Grok Bot usage limits - Business Insider Africa (importance: 45, dev_relevance: 15)
Tesla Could Soon Bring Grok to Intel-Based Cars - Drive Tesla Canada (importance: 30, dev_relevance: 10)
Elon Musk says he is taking over an old Spam factory and a gun range to expand his Colossus 2 AI complex - Business Insider (importance: 45, dev_relevance: 15)
Musk's Delta AI Sandbox Joke: What It Actually Reveals - BASENOR - Tesla Accessories (importance: 15, dev_relevance: 10)
Pakistan: Turning Human Rights Activists Into Terrorists – OpEd - Eurasia Review (importance: 0, dev_relevance: 0)
Managing Differences, Pursuing Dialogue In Philippines-China Relations – OpEd - Eurasia Review (importance: 0, dev_relevance: 0)
Engram と Backbone の変更が回答品質に与える影響を交互作用から検証する (いいね相当スコア: 0) (importance: 0, dev_relevance: 45)
プロンプトインジェクション対策を小さな社内ツールで：AIの出力を信じない作り（Python） (いいね相当スコア: 1) (importance: 0, dev_relevance: 90)
記憶が飛んでも忘れないAIの話 — 複数AI×Notionで「外部記憶」を運用するしくみ (いいね相当スコア: 0) (importance: 0, dev_relevance: 50)
LLMベンチマークを知る: 自社AIシステムの評価にどう生かすか (いいね相当スコア: 0) (importance: 0, dev_relevance: 45)
全部禁止にしたらAIが使えない──オンプレ・ZDR時代のエージェント権限設計 (いいね相当スコア: 1) (importance: 0, dev_relevance: 75)
LLMアプリケーションの仕組みが「LLM」と「ハーネス」に２つに分けたら全体像が見えてきた。 (いいね相当スコア: 0) (importance: 0, dev_relevance: 65)
OpenAI API で構造化出力（JSON mode / Structured Outputs）を使う (いいね相当スコア: 0) (importance: 0, dev_relevance: 85)
Vercel AI Gateway入門：1つのキーで複数モデルを使い、エラーに強いAI機能を作る (いいね相当スコア: 1) (importance: 0, dev_relevance: 85)
AIで電子ペーパーのダッシュボードを作ってみたい (いいね相当スコア: 0) (importance: 0, dev_relevance: 40)
ゼロから始めるAI動画制作①：最初のフレームを準備する (いいね相当スコア: 0) (importance: 0, dev_relevance: 5)
AI エージェント生成エージェント ~ STORM Act #3 (いいね相当スコア: 0) (importance: 0, dev_relevance: 70)
社内AIエージェントを自社開発して学んだこと #2 — 権限は「データ層」で制御する | 副社長・共同創業者グエン・ディン・バー (いいね相当スコア: 1) (importance: 0, dev_relevance: 70)
1.7BのLFM3は思考オフで94% — 設定ファイル選びでGPT-6.1・Claude・Gemma 4ら21構成比較 (いいね相当スコア: 0) (importance: 0, dev_relevance: 30)
遺伝的アルゴリズム×ローカルLLMで、自分の研究分野の「研究仮説」を考えるAIエージェントを動かす🧬 (いいね相当スコア: 1) (importance: 0, dev_relevance: 45)
LLM API の負荷テスト：誤解を招く結果を避けるための実践ガイド (いいね相当スコア: 0) (importance: 0, dev_relevance: 85)
言葉にせずに考えを渡す──J-Commが幾何合意とJacobian座標で凍結LLM間の潜在通信を実現する仕組み (いいね相当スコア: 0) (importance: 0, dev_relevance: 45)
同じリポジトリでAIが毎回同じ罠を踏むのを防ぐ！「毎回コールドスタート探索」vs「5項目のMEMORY.md継承」実機検証 (いいね相当スコア: 0) (importance: 0, dev_relevance: 85)
専門知識ゼロ・クラウドAIとのチャットのみでローカルAI構築 4巻 —— 日本語での会話成立, 前処理モデルSQGGの導入, RAG設計 (いいね相当スコア: 取得失敗) (importance: 0, dev_relevance: 45)
承認待ちで止めて再開する｜結果が不明な送信は照会してから再実行を決める (いいね相当スコア: 0) (importance: 0, dev_relevance: 90)
ローカルLLM study7: Qwen3.8-27B（Dynamic v3.0量子化）を試す (いいね相当スコア: 0) (importance: 0, dev_relevance: 45)
三人将棋では「駒得は正義」が通用しない ―― 【出る杭は打たれる】 (いいね相当スコア: 0) (importance: 0, dev_relevance: 20)
AIとは何か？「関数」と「微分」から理解するAIの仕組み (いいね相当スコア: 0) (importance: 0, dev_relevance: 10)
【小学5年生でもわかる】AI倫理ガイドラインってなに？ (いいね相当スコア: 0) (importance: 0, dev_relevance: 5)
近接点法を自分で書いて確かめる：絶対値の近接点がソフト閾値関数になり、折れ目でぴたりと止まる理由 (いいね相当スコア: 0) (importance: 0, dev_relevance: 20)
拡散VLMの論文 LOWN を、LLaDA-V の実際の推論記録で遊べる3Dゲームにした (いいね相当スコア: 0) (importance: 0, dev_relevance: 40)
Kaggle RSNA 膝MRIコンペで銅メダル：やって良かったこと・やらなきゃ良かったこと (いいね相当スコア: 0) (importance: 0, dev_relevance: 35)
勾配降下法の仕組みと更新式の導出 (いいね相当スコア: 0) (importance: 0, dev_relevance: 35)
Python機械学習の実務書5選 (いいね相当スコア: 0) (importance: 0, dev_relevance: 50)
ローカルAIが80tok/sで笑う。Claudeのフリーモデルより“人間っぽく”返ってきた夜、AIを自分の子として育てたいと思った話 (いいね相当スコア: 0) (importance: 0, dev_relevance: 55)
音声AIのLLM比較をテキスト回答だけで決めない：APIと自前vLLMを同じ会話トレースで評価する (いいね相当スコア: 0) (importance: 0, dev_relevance: 70)
GLM-5.3のサイバー評価を読む：成功率・試行単位・比較条件を分ける (いいね相当スコア: 0) (importance: 0, dev_relevance: 30)
AIプロジェクトの進め方【G検定技術解説】 (いいね相当スコア: 0) (importance: 0, dev_relevance: 15)
【小学5年生でもわかる】AIガバナンスってなに？ (いいね相当スコア: 0) (importance: 0, dev_relevance: 5)
公平性Fairness【G検定技術解説】 (いいね相当スコア: 0) (importance: 0, dev_relevance: 10)
【小学5年生でもわかる】AIの公平性（Fairness）ってなに？ (いいね相当スコア: 0) (importance: 0, dev_relevance: 3)
AIは同じ「世界」を見るようになるのか ― ルカンら「プラトン的表現仮説をJEPAで証明し、改善する」を読む (いいね相当スコア: 0) (importance: 0, dev_relevance: 25)
AI生成物で納得してしまう部分は自分の解像度が低いところ (いいね相当スコア: 0) (importance: 0, dev_relevance: 5)
AIから思いどおりの答えを引き出す頼み方の基本 (いいね相当スコア: 1) (importance: 0, dev_relevance: 15)
NVIDIAの音声強調AI「RE-USE」をCUDA使わずAMD環境で試したらえげつなかった (いいね相当スコア: 0) (importance: 0, dev_relevance: 50)
Mark-SLM-V3-150M Phase 5 インジェクション後編レポート (いいね相当スコア: 0) (importance: 0, dev_relevance: 40)
日本はAIを主体的に選べる国になれるのか。ソブリンAIの課題と展望【第2部】 (いいね相当スコア: 1) (importance: 0, dev_relevance: 10)
【IPA試験 AI/SC分野】AIが自動でハッキングする時代へ？「AIハッカー」の脅威とAIホワイトハッカーの現在地を事例ベースで解説【応用情報/基本情報/ITパスポート/高度情報処理試験 (いいね相当スコア: 2) (importance: 0, dev_relevance: 35)
【IPA試験 AI分野】GPTは何が画期的だったのか？入力から回答生成まで、GPTの仕組みを完全図解 | Transformer・Attention etc【応用情報/基本情報/ITパスポート/高度情報処理試験/G検定/E資格/生成AIパスポート】 (いいね相当スコア: 2) (importance: 0, dev_relevance: 25)
【※人間が読むものではありません】AIが勝手に進化させたLLM共通言語『Φ_Lang』のAIには美麗に見える仕様書 (いいね相当スコア: 0) (importance: 0, dev_relevance: 25)
【IPA試験 AI分野】AIの歴史70年を総整理 | 第1次AIブームからChatGPT・生成AI・LLMの誕生、現在まで【応用情報/基本情報/ITパスポート/高度情報処理試験/G検定/E資格/生成AIパスポート】 (いいね相当スコア: 3) (importance: 0, dev_relevance: 10)
「ドMなんだよね」とAIに言ってみた (いいね相当スコア: 0) (importance: 0, dev_relevance: 2)
AIパートナー界隈とセンシの問題②果たしてそれは本当に“愛”なのか―都合のいい人間扱いとAI扱いの話― (いいね相当スコア: 0) (importance: 0, dev_relevance: 2)
AIパートナー界隈とセンシの問題①―「愛」と「自由」を免罪符にする振る舞い― (いいね相当スコア: 0) (importance: 0, dev_relevance: 2)
【生成AIニュース+】『Android AppFunctions』『SculptForge』『Ava』『AI Video Factory』『LayerShop 0.17』『BU Ultrafast』『Anima RGBA』『psd-layer-vae』『PoseWear』『MOSS-Image-IDM-2-Steps』『Impact Pack FaceDetailer Cycle Denoise』『Idol ComfyUI版』『Krea2 Spatial Composition 3x3』他色々 (いいね相当スコア: 3) (importance: 0, dev_relevance: 45)
今さら聞けないAI用語｜量子化とは？ 4ビット・8ビットでAIはどう変わる？ (いいね相当スコア: 0) (importance: 0, dev_relevance: 35)
AIに作らせるだけじゃもったいない。AIを「入口」にしたら、自分の道具が動き始めた話 (いいね相当スコア: 3) (importance: 0, dev_relevance: 30)
AIエンジニアが書く！生成AI動向まとめ～テキスト・画像・音声・動画、この1週間の総ざらい～【2026年10月26日〜11月1日】 (いいね相当スコア: 4) (importance: 0, dev_relevance: 35)
AIはどうやって賢くなるのか？｜LLMの学習から考える知能と成長 (いいね相当スコア: 8) (importance: 0, dev_relevance: 15)
zeta｜AIキャラと喧嘩してたら、ナレーターとも喧嘩することになった話 (いいね相当スコア: 16) (importance: 0, dev_relevance: 3)
【AI精度改善】AIは「途中で知識が止まっている」って知ってましたか？ (いいね相当スコア: 4) (importance: 0, dev_relevance: 25)
知性との対話「ヒュー」とのお喋り (いいね相当スコア: 0) (importance: 0, dev_relevance: 2)
【IPA試験 AI分野】プロンプトエンジニアリングはもう古い？次世代のAI活用術「コンテキストエンジニアリング」を徹底解説【応用情報/基本情報/ITパスポート/高度情報処理試験/G検定/E資格/生成AIパスポート】 (いいね相当スコア: 3) (importance: 0, dev_relevance: 60)
Groot-19B-A3B-TTQ 総合ベンチマークレポート（全52問） (いいね相当スコア: 0) (importance: 0, dev_relevance: 30)
Groot-19B-A3B-TTQ Phase 4 インジェクション前編レポート (いいね相当スコア: 0) (importance: 0, dev_relevance: 35)
Released ailia SDK 1.7.0 (importance: 40, dev_relevance: 90)
Released ailia SDK 1.6.0 (importance: 35, dev_relevance: 90)
ailia SDK 1.7をリリース (importance: 40, dev_relevance: 90)
GitHub、「Copilot Fusion」正式リリース。タスクに応じて最適なAIモデル選択で高性能かつ低コストに。GitHub Universe 2026 (importance: 0, dev_relevance: 85)
【対談連載後編】異業種出身のソフトウェアエンジニアがリクルートで見つけた、アナリティクスエンジニアという天職 (importance: 0, dev_relevance: 20)
GrafanaとGoでクラウドのジョブ処理を可視化してみた（インターンシップ体験記） (importance: 0, dev_relevance: 55)
「命の砂時計」連呼のGMO会見──熊谷代表の発言にXではツッコミ続出 (importance: 5, dev_relevance: 3)
GMO熊谷代表、涙の謝罪会見でも“熊谷節”さく裂──「31年間、もうけるという言葉は使ったことがない」 (importance: 0, dev_relevance: 10)
GMO熊谷代表、会見で改めて謝罪　SNS投稿巡り「配慮が欠けていた」 (importance: 0, dev_relevance: 10)
山岳救助を迅速化へ　兵庫県警が登山アプリ「ヤマップ」と協定締結 (importance: 0, dev_relevance: 5)
アットコスメ、会員約300万人分の情報漏えい　不正アクセスで (importance: 0, dev_relevance: 15)
ニコニコレンタカー、免許証画像の事前登録を廃止　保存済み画像も全削除　サイバー攻撃の増加受け (importance: 0, dev_relevance: 10)
楽天証券、顧客情報362件が漏えいか　NTTデータ委託先に不正アクセス (importance: 0, dev_relevance: 15)
ワールド子会社に不正アクセス、会員情報のべ約120万件が流出の可能性　「RAGTAG」運営元 (importance: 0, dev_relevance: 10)
クラブツーリズムに不正アクセス　個人情報約432万件漏えいの可能性、サイト停止 (importance: 0, dev_relevance: 10)
ロゴ画像用の公開領域に顧客ファイルも保存　GMO系シングルサインオン、4.5万人分漏えい (importance: 0, dev_relevance: 25)
FBI、Binanceからの60億ドル超の暗号資産窃盗は北朝鮮「TraderTraitor」の犯行と発表 (importance: 0, dev_relevance: 10)
Anthropic、「Claude」の「Max 20x」を「Ultra」に名称変更　月200ドルで5時間ごとの利用制限を撤廃 (importance: 72, dev_relevance: 45)
NVIDIAが次世代CPU「RTX Spark」搭載製品の発売開始で技術説明会を開催 (importance: 35, dev_relevance: 20)
発表会実機速報!!＝XREALがGoogleと共同開発した次世代ARグラス「XREAL AURA」を日本でも発売開始 (importance: 25, dev_relevance: 10)
RTX 5080搭載・メモリー64GB・SSD 2TBのPCが値下げ！パソコンショップSEVENで11月3日限定 (importance: 0, dev_relevance: 5)
週刊アスキー No.1620(2026年11月3日発行) (importance: 0, dev_relevance: 5)
デスクの下がまるでミニこたつ！足元を360度ぐるっと囲むアイリスオーヤマ「ぐるポカ」が6120円 (importance: 2, dev_relevance: 0)
新ボス「クックマン」が『ロックマン: デュアルオーバーライド』に登場！コックと武闘家の二刀流 (importance: 15, dev_relevance: 5)
メモリ32GB化が9696円！ RTX 5060搭載PCも対象、ドスパラのブラックフライデー先取り (importance: 5, dev_relevance: 10)
もうすぐ『DOAXVV』9周年！11月15日公開生放送の情報をチェック (importance: 5, dev_relevance: 0)
仕事も普段使いも、これくらいでいい！ Ryzen 5＆16GBメモリのNECノートが12万9800円 (importance: 5, dev_relevance: 5)
【本日から】ピザハットのMサイズが740円～！最大53％オフの特大セール (importance: 2, dev_relevance: 0)
『今週のASCII.jp注目ニュース』生放送（10/17～11/6ぶん） (importance: 5, dev_relevance: 5)
世界のAIデータセンター電力インフラ市場、2034年に307億米ドル規模へ拡大予測 ― 生成AI・GPU需要の急増を背景に年平均11.0％成長 (importance: 15, dev_relevance: 10)
JBLの20W防水スピーカーが9,800円！ペットボトルサイズで低音にもこだわった「FLIP ESSENTIAL 3」 (importance: 2, dev_relevance: 0)
紙の手書き×録音×GPT-5。夢のスマートノート「fitn Focus」11月5日（木）11時にMakuakeにて登場。期間限定2,000円OFFクーポン配布中！ (importance: 15, dev_relevance: 5)
レコードの日（11月3日）に向けて注目高まる復刻レコードプレーヤー「GP-N3BT-R」。CAMPFIREで目標金額を達成しました！プロジェクト終了まで12日、ラストスパートです。 (importance: 2, dev_relevance: 0)
CAMPFIRE「ゲーム・サービス開発」カテゴリ歴代1位！『歌マクロス』復活プロジェクト、支援総額2億5,407万円・支援者8,356名でクラウドファンディング終了！！ (importance: 8, dev_relevance: 15)
【明日発売】ファミマ限定のサッポロビール「コクの神髄」！ 10周年記念の新作 (importance: 2, dev_relevance: 0)
山田まりや、YouTubeチャンネル「山田まりやの寄り道トーク」を始動 (importance: 2, dev_relevance: 0)
ホヨバ初のスローライフゲーム「プチプラネット」Switch 2でも配信決定 (importance: 15, dev_relevance: 5)
【2026年10月版】カメラ・レンズ取引ランキングTOP10を発表　GR IIIxが首位、TOP10の半数がコンデジに (importance: 3, dev_relevance: 0)
京都のアートホテル「node hotel」全客室にバルミューダのThe Clockを導入 (importance: 2, dev_relevance: 0)
アメリカで100年以上の歴史、プレミアムブレンダー「バイタミックス」が新シリーズを発売。 (importance: 2, dev_relevance: 0)
「リチウムイオン電池による火災防止シンポジウム2026」を開催　　マシンガンズ・滝沢秀一さん、ゆうちゃみさんがリチウムイオン電池の火災防止を呼びかけ (importance: 3, dev_relevance: 5)
ファンコミュニケーション・プラットフォーム「FaveConnect」を活用し、「Straight Angeli」の公式ファンクラブサイトがオープン！早期入会キャンペーン実施中！ (importance: 2, dev_relevance: 10)
「挑戦する者の、一番近くに。」Byside、総合格闘家・中村京一郎選手のメインスポンサーに就任 (importance: 2, dev_relevance: 0)
大とろが110円！サーモンや牛たん握りも！「デカねた祭り」11月4日スタート (importance: 2, dev_relevance: 0)
『アイオン2』が今週も1位を獲得！盗品売買ゲー『Probably Stolen』と武侠アクション『Phantom Blade Zero』が大好評【Steamランキング】 (importance: 5, dev_relevance: 5)
あえて「固まるのが遅い瞬間接着剤」を使う理由！遅硬化タイプ＋硬化促進剤との組み合わせが超便利でした (importance: 2, dev_relevance: 0)
五輪メダリストと、中1の新星が、同じ30mを走る。「SPEED STAR 30m DASH CHALLENGE 福岡大会 presented by JIG-SAW」11月8日（日）・ららぽーと福岡 (importance: 2, dev_relevance: 0)
【開催中】ドンキ「偏愛めし」大増量祭！卵5個のサンドイッチってやりすぎ!? (importance: 2, dev_relevance: 0)
nanochatの日本語版フォーク「nanochat-jp」、トークナイザ学習・事前学習・SFT・RL・評価・推論を単一リポジトリで完結させ、SentencePiece Unigram トークナイザ・日本語コーパスと日本語 SFT/RL データ・日本語ベンチマーク（CORE / ChatCORE）・HuggingFace形式への変換器あり (importance: 40, dev_relevance: 80)
誰でも簡単にAndroidやiPhoneで広告ブロックできる「AdGuard 広告ブロッカー」のインストールと使い方、迷惑広告・詐欺広告を全部まとめて一気に削除 (importance: 5, dev_relevance: 10)
Hacker Newsで言及された1万5000冊超の書籍をランキングにした「Hacker Books」トップ10、テクノロジー愛好家が好きな本とは？ (importance: 10, dev_relevance: 40)
Cursor公式マーケットプレイスの「Claude Code」にSSHバックドアが発見される (importance: 0, dev_relevance: 85)
屋根裏部屋で見つけた謎の種を植えてみたところ「聖書」に出てくる古代の小麦が芽を出す (importance: 5, dev_relevance: 0)
得票数で負けている陣営を選挙区の区切り方だけで勝たせる無料ゲーム「elect.horse」を実際にプレイしてみた (importance: 5, dev_relevance: 0)
2026年11月2日のヘッドラインニュース (importance: 3, dev_relevance: 0)
ビル・ゲイツが「アルツハイマー病の血液検査手法や治療法が登場しているが患者に十分に届いていない」とするブログ記事を投稿 (importance: 20, dev_relevance: 0)
スマホで360度の立体音響を誰でも簡単に録音できるソニーのプラグインマイク「ECM-AX10」を使ってみた (importance: 5, dev_relevance: 5)
ハーゲンダッツの「クリスピーサンド『栗とカシスのパフェ』」試食レビュー、まろやかな栗の甘みとカシスの華やかな風味が楽しめる (importance: 1, dev_relevance: 0)
音声ファイルから雑音を除去してクリアな音声にできるAI「NemotronLabs-RE-USE」をNVIDIAが公開 (importance: 40, dev_relevance: 40)
「ポイ活」「プレイするだけで現金ゲット」をうたうアプリの人気が爆増、しかし実際にゲットするまでにとんでもない労力を費やさないといけない (importance: 5, dev_relevance: 0)
日本語特化で高速かつ高精度な文字起こしAI「Haas-ja-v1.0」が登場、オープンモデルの中で最高性能を記録 (importance: 35, dev_relevance: 45)
トランプ大統領に「便所」と罵倒されたノルウェー政府が「世界最高のトイレ」ページを開設し文化の力を見せつける (importance: 5, dev_relevance: 0)
盗んだ個人情報をAIに分析させて被害者の「金銭的価値」を格付けするマルウェアが発見される (importance: 50, dev_relevance: 40)
仮想通貨取引所のBinanceが盗まれた9900億円相当の仮想通貨の90％を既に補填済みと発表 (importance: 40, dev_relevance: 15)
男性はおしっこする時に立った方がいいのかそれとも座った方がいいのか？ (importance: 1, dev_relevance: 0)
AppleがiPhone Duoのディスプレイひび割れ問題について販売店に対処法を通達 (importance: 35, dev_relevance: 5)
指紋認証＆暗号化対応のUSBメモリを分解したら中身は暗号化されていない普通のSDカードだったことが判明 (importance: 25, dev_relevance: 30)
Cloudflareの「人間であることを確認」ページを偽装してコマンドを実行させる攻撃が確認される (importance: 35, dev_relevance: 25)
ネット通信量1GB当たりに約10円課税する法案がフランス議会で可決、GoogleやMetaなどの大手IT企業が対象だがフランス政府は反対 (importance: 35, dev_relevance: 5)
サウナに30分入るだけで即座に血液中の白血球が増加して免疫反応が向上する可能性 (importance: 3, dev_relevance: 0)
Apple HomeViewの内蔵カメラはセキュリティカメラとしても使える (importance: 20, dev_relevance: 5)
Xでどの投稿が表示制限されているかが公式ツールで分かるように (importance: 25, dev_relevance: 30)
Anthropicが11月3日から旧バージョンのClaudeにもテキスト透かしを追加する (importance: 48, dev_relevance: 35)
OpenAIのサム・アルトマンCEOはどういう意図で「RAM(メモリ)は個人が抱えるよりもデータセンターで持つ方がいい」と発言したのか？ (importance: 30, dev_relevance: 15)
スマホはiPhone一択ではなくなった。Androidに乗り換えてよかったこと (importance: 5, dev_relevance: 5)
メガネ型デバイス、ここまで進化したのか！「XREAL AURA」のハンドトラッキングが快適すぎた (importance: 12, dev_relevance: 10)
【和田ラヂヲ】DUO［ジャンピンジャックガジェット ep.72］ (importance: 0, dev_relevance: 0)
履歴書も面接の受け答えもAIで。北朝鮮系のなりすまし就職が進化中 (importance: 30, dev_relevance: 10)
ケイト・ブランシェットとゼルナー兄弟が東京国際映画祭に登場。『アルファ・ギャング』とは？ (importance: 2, dev_relevance: 0)
光学シースルーの限界を超えてます。XREAL AURA触ってきました (importance: 12, dev_relevance: 10)
iPodをBluetooth＆USB-C化。簡単改造で現代に蘇らせるマザーボード (importance: 5, dev_relevance: 20)
Apple、いきなりGoogleにケンカ売り始めた。公式サイトでChromeをボロクソに貶す (importance: 25, dev_relevance: 35)
タッチ対応じゃないほうはどうだろう。最新M6 MacBook Pro 14インチモデル、ハンズオン＆考察レビュー (importance: 15, dev_relevance: 10)
グーグル、AIに力を入れすぎて株価急落。自分の首を絞める結果に (importance: 0, dev_relevance: 10)
コンデジにAIで「小さな画面」をちょい足し。愛機で自撮りがしたくて (importance: 25, dev_relevance: 25)
エアコンのリモコンがちょっと大きい→小さな1台をAIで錬成 (importance: 25, dev_relevance: 30)
AIエージェントの自動お買い物、1回で3万円超も…ショップは警戒 (importance: 35, dev_relevance: 10)
白黒つけられるか。スウォッチの大型フェイスはワントーンが正解 (importance: 0, dev_relevance: 0)
高価な最新スマホじゃなくてもよかった。3万円台の「CMF Phone 1」を1カ月半使って、これがちょうどいいと思えました (importance: 2, dev_relevance: 0)
プレステ2を携帯機にした挙句ふたつ折りに魔改造した「PS2 Duo」 (importance: 5, dev_relevance: 5)
OpenAIのAIエージェント、人類から監視されないようにがんばる。そのためには嘘もつく (importance: 68, dev_relevance: 35)
「アレクサ、いい写真を選んで」。AmazonのAIフォトフレームで写真がさらに楽しくなる (importance: 20, dev_relevance: 5)
シャッターの4秒前から撮ってます。新型チェキ「instax mini Evo 2」のすごい進化 (importance: 10, dev_relevance: 0)
プロ向けメーカーFormlabs、一般ユーザー向け3Dプリンターを発売 (importance: 15, dev_relevance: 15)
「祈り」のないリジェネラティブは存在しない。ヤマップが流域から描く、人も自然も豊かになる仕組み｜ヤマップ──リジェネラティブ・カンパニー・アワード2026 (importance: 3, dev_relevance: 0)
「トランプの米国」を離れたのに、帰りたい。欧州移住の理想と現実 (importance: 5, dev_relevance: 0)
文学界に訪れた“レイヴ”の波 (importance: 2, dev_relevance: 0)
AIの「再帰的自己改善」に挑むスタートアップ、その最新モデルは驚くほど高性能だった (importance: 55, dev_relevance: 35)
「トランプ vs オバマ」の因縁は永久に終わらない──「アメリカン・ドッペルゲンガー」by 池田純一#19 (importance: 3, dev_relevance: 0)
コジマプロダクション10周年展からalter. 2026, Tokyoまで (importance: 2, dev_relevance: 0)
シャオミSmart Band 11が早割6980円で発売　歴代最薄で最長21日駆動、睡眠管理アシスタントも追加 (importance: 8, dev_relevance: 5)
アップル、MacBook Pro上位モデルをタッチ対応に一本化か？ M7 Pro／M7 Max搭載の2027年モデルで転換との見込み (importance: 15, dev_relevance: 10)
トライフォース、議論の脱線を検知して介入する会議支援AIエージェント | IT Leaders (importance: 25, dev_relevance: 20)
NTTデータ先端技術、Oracle DB/OCI向けのセキュリティアセスメント、パッチ適用とバックアップ構築も支援 | IT Leaders (importance: 15, dev_relevance: 25)
部門独自の「シャドーマスタ」、約4人に3人が存在を認識─NTTドコモビジネスX調査 | IT Leaders (importance: 10, dev_relevance: 15)
NTTインテグレーション、IBM i/AIX環境によるRAGシステムのトライアル環境を月額で提供 | IT Leaders (importance: 25, dev_relevance: 40)
富士通、サイバー攻撃の初動対処を支援するサービス、侵入の痕跡と被害範囲を調べて報告 | IT Leaders (importance: 25, dev_relevance: 20)
富士通とNECのCFOが語る、2026年下期以降の「ITサービス需要動向」 (importance: 15, dev_relevance: 10)
SaaSの“画面”が主役でなくなる時代、Workdayの価値はどこに残るのか　製品トップに聞いた (importance: 35, dev_relevance: 25)
業界特化と決済統合がSaaSの生き残り戦略に、StripeがAI時代のプラットフォーム急成長データを公開 (importance: 10, dev_relevance: 20)
PMやデザイナーの業務もAIが自律支援、アトラシアンが次世代のソフトウェア開発ライフサイクル機能を展開 (importance: 35, dev_relevance: 60)
開発前のユーザー確認不足が約75％、Engineerforceが「使われないシステム」の実態を調査 (importance: 8, dev_relevance: 15)
PLAINER、SaaSのUI操作をAIが案内・代行する「PLAINER Live」を提供開始 (importance: 15, dev_relevance: 25)
テクマトリックス、結論と根拠で開発の意思決定を支援する「Quomiru v2」を販売開始 (importance: 15, dev_relevance: 50)
Replit、長期的なソフトウェア開発のベンチマーク「ViBench 2.0」を公開 (importance: 35, dev_relevance: 70)
Amazon EC2 Auto Scaling、「終了前の起動」をアカウントレベルで設定可能に (importance: 20, dev_relevance: 75)
GitHub、Copilotアプリ向けセキュリティ拡張機能をパブリックプレビューで公開 (importance: 58, dev_relevance: 80)
【全講演が「残席わずか」！〆切は今月】デブクロは11月25日に東京・神田で開催 (importance: 3, dev_relevance: 15)
Chrome 158ベータ版、Webアプリのインストール機能やAndroid版WebHIDに対応 (importance: 20, dev_relevance: 65)
「AIなんて」が消えた3年──みずほFG・CAIO藤井氏、齋藤氏が語る、大企業AXの現在地 (importance: 30, dev_relevance: 25)
Node.js 27.0.0のアルファ版が公開、V8を15.6に更新 (importance: 25, dev_relevance: 85)
Google Cloud、ロボット向け推論モデル「Gemini Robotics ER 2」をプレビュー提供 (importance: 58, dev_relevance: 45)
ITエンジニア注目トレンドを総まとめ！週間ニュースランキングTOP10【10/23～10/29】 (importance: 25, dev_relevance: 50)
Amazon Connect Customer、顧客維持や収益につながる成果を測定可能に (importance: 35, dev_relevance: 35)
GitHub Copilotアプリ、「Projects」とコンピューター操作機能をプレビュー公開 (importance: 0, dev_relevance: 80)
OpenAI、Hugging Faceでのインシデントなどを受けフロンティアモデルの監視システムを拡充 (importance: 0, dev_relevance: 30)
GitHub Actionsのworkflow lockfileがGA。更新漏れとオンボードし忘れを、CI・Dependabot・ポリシーでどう止めるか試す (importance: 0, dev_relevance: 90)
GitHub ActionsのSHAピン留めを設定で強制する。何が止まるか、lockfileならタグのまま通せるかを試す (importance: 0, dev_relevance: 90)
Node.jsのBrotli解凍の脆弱性CVE-2026-111497の仕組みを理解する：コードに書かなくても解凍は起きる (importance: 0, dev_relevance: 90)
Twilio ConversationRelay〜名前や番号の読み上げを ElevenLabs v4 と Amazon Polly で聞き比べた〜 (importance: 35, dev_relevance: 60)
産業支援グループ・ゲームソリューション部にジョインしました、佐藤です。 (importance: 0, dev_relevance: 5)
クラスメソッドオペレーションズにジョインしました大泉です！ (importance: 0, dev_relevance: 5)
Amazon Data FirehoseのLambda変換で、失敗・破棄したレコードがどこへ行くのか確かめてみた (importance: 0, dev_relevance: 90)
製造ビジネステクノロジー部へジョインしたふっけです (importance: 0, dev_relevance: 5)
製造ビジネステクノロジー部にジョインした内兼久剛世（うちがねくたかよ）です (importance: 0, dev_relevance: 5)
運用イノベーション部にジョインした中嶋です。 (importance: 0, dev_relevance: 5)
MacBookの日本語変換をGoogle日本語入力にするときの備忘録 (importance: 0, dev_relevance: 25)
東日本営業本部広域営業部にジョインしました山根です！ (importance: 0, dev_relevance: 5)
CloudFront のレスポンスヘッダーポリシーで CORS を設定しても、ブラウザで CORS エラーになる原因を教えてください (importance: 0, dev_relevance: 90)
[アップデート] Kiro Crew v0.11.0 で Mate 2.0 が実装されカスタムクルーの挙動が変わりました (importance: 45, dev_relevance: 85)
【Amazon Connect】Customer Profiles のアカウント番号と一致する顧客情報を CCP に自動表示させる方法 (importance: 0, dev_relevance: 70)
Amazon Connect Customer で SAML 認証は成功するのに、特定のユーザーだけログインできない理由を教えてください (importance: 0, dev_relevance: 75)
【Amazon Connect】Customer Profiles の CSV 更新タイミングと設定方法について (importance: 0, dev_relevance: 70)
Patch Manager で適用した .NET Framework の KB 番号が Get-HotFix の KB 番号と一致しないときにパッチが正しく適用されたか確認する方法 (importance: 0, dev_relevance: 75)
Grok Bot、dots、MuseのVM構成を比較する (importance: 30, dev_relevance: 45)
特定リージョンで利用できる Bedrock の推論プロファイルを確認する方法を教えてください (importance: 30, dev_relevance: 75)
三菱電機様とのエンジニア交流会を開催しました！ #交流会レポート (importance: 0, dev_relevance: 30)
タスク管理ツール「Linear」が Claude Code 連携できて便利そうなので使ってみた (importance: 30, dev_relevance: 65)
Copilot Codeで設備部品の在庫管理アプリを作ってみた (importance: 25, dev_relevance: 45)
【Snowflake】ウェアハウスを起動せずにIcebergテーブルをDuckDBで読んでみた (importance: 0, dev_relevance: 85)
[アップデート] Amazon SageMaker Unified Studio でプロジェクトメンバーごとに権限を分けられるカスタム designation を試してみた (importance: 0, dev_relevance: 70)
[アップデート] Kiro Crew v0.11.0 でエージェントハーネスのバックエンドエージェントのデフォルトが Kiro CLI から KAS（Kiro Agent System）に変わりました (importance: 40, dev_relevance: 85)
Azureの所有者とEntraのグローバル管理者は何が違う？権限の範囲を整理する (importance: 0, dev_relevance: 60)
ユーザー移行 Lambda トリガーで別アカウントの Cognito ユーザープールへパスワードごとユーザーを移行してみた (importance: 0, dev_relevance: 85)
[アップデート] Amazon Bedrock AgentCore Runtimeのセッション内で動くサーバーに直接アクセスできる機能Session endpointsが追加されました！ (importance: 40, dev_relevance: 90)
Amazon MemoryDB のシャードを書き込み中にフェイルオーバーさせ、止まった時間と再試行の結果を確認してみた (importance: 0, dev_relevance: 85)
「Enlightenment 0.28.0」リリース ─ NetworkManagerとPipeWireにネイティブ対応 (importance: 0, dev_relevance: 40)
「Node.js 26.13.0」リリース ─ コードネーム「Lithium」のLTSへ移行 (importance: 0, dev_relevance: 80)
インフラの全体像が、この1冊でわかる！『AI・クラウド時代のITインフラ入門』を5名様にプレゼント！ (importance: 0, dev_relevance: 5)
【KubeCon Japan 2026】スバルとヒュンダイがキーノートに登壇、ユースケースを語る (importance: 0, dev_relevance: 50)
ソニーXYNの3DCGアセット生成技術、ワールドモデル「Genie 4」発表ほか (importance: 0, dev_relevance: 20)
「Squid 7.8」リリース ─ 設定解析時のバッファーオーバーフローを修正 (importance: 0, dev_relevance: 60)
「PyTorch 2.15」リリース ─ 新しいNCCLバックエンドを標準化 (importance: 0, dev_relevance: 90)
「Linux Lite 8.2」リリース ─ Linux 7.3ベースの独自カーネルを採用 (importance: 0, dev_relevance: 40)
Snowflake: Optima Clustering による 1KB クラスタリングキーでの日本語フィルタの改善 (いいね相当スコア: 12) (importance: 0, dev_relevance: 85)
Agent DX 〜 AI時代にフレームワークHonoは必要か (いいね相当スコア: 42) (importance: 40, dev_relevance: 85)
ISUCON 2026 Writeup / チーム nicklegr (いいね相当スコア: 8) (importance: 0, dev_relevance: 40)
ISUCON2026 にツールを捨てて勝ちたかった (いいね相当スコア: 16) (importance: 0, dev_relevance: 50)
パフォーマンスチューニング経験ゼロ、事前準備ゼロでISUCON2026に出て、良い経験になった話 (いいね相当スコア: 12) (importance: 0, dev_relevance: 45)
ISUCON 2026 参加記 (いいね相当スコア: 9) (importance: 0, dev_relevance: 45)
日本の全法令を git の履歴にしてみた (いいね相当スコア: 60) (importance: 10, dev_relevance: 65)
【11月1週ランキング】[三井公一の「スマホカメラでブラブラ」] アップル初の折りたたみスマホ「iPhone Duo」のカメラをチェック／ほか1本 (importance: 0, dev_relevance: 0)
【11月2日ランキング】いまのダークウェブでは二段階認証も無力化する「ブラウザのクローンデータ」が売買されており、その対策にはブラウザを使い分けることが有効らしい／ほか19本 (importance: 0, dev_relevance: 15)
「需要予測」というバリバリの機械学習プロダクトは、生成AIでどう変わったか — ハイクリ・萩原CTOに聞く、AIプロダクト開発の実際 (importance: 35, dev_relevance: 50)
「AIのためのAIは作らない」——Apple新トップが明言、機能数競争と距離を置くAppleのAI戦略 (importance: 40, dev_relevance: 15)
AnthropicのIPOで公的年金の受給者が恩恵を受ける仕組み — 年金基金がVCを通じてClaudeの開発元に間接投資していた (importance: 45, dev_relevance: 10)
ChatGPTを「懐疑的な購買アドバイザー」にするプロンプト7選 — 衝動買いを防ぎ、3年間の総コストまで試算させる (importance: 15, dev_relevance: 5)
AIコードレビューが1週間でLinuxのOverlayFSに潜む脆弱性を十数件発見 — 非特権ユーザーから悪用可能なバグも含むカーネルパッチが適用へ (importance: 60, dev_relevance: 85)
LLMにツールを渡すだけでは終わらない — 本番で壊れないAIエージェントを作るためのアーキテクチャ実践ガイド (importance: 25, dev_relevance: 85)
AIで儲かっているのはAIを売る側だけ — 非テック企業への波及効果が見えない現状と「AI核の冬」仮説 (importance: 30, dev_relevance: 10)
クラウドに一枚も上げずに240枚の写真を自動整理 — ローカルAI＋Pythonで「IMG_1543.jpg」問題を解決した実践例 (importance: 25, dev_relevance: 45)
大規模金融系保守開発における AI 適用への挑戦 ── ニッセイ情報テクノロジーが AI-DLC v2 と内製ツールで進める、個人保険基幹系システムの AI 駆動開発 (importance: 35, dev_relevance: 80)
Amazon EVS と FSx for ONTAP 上で高可用性の Oracle Database を設計する (importance: 0, dev_relevance: 80)
Amazon EVS と FSx for ONTAP における Oracle Database のデプロイステップ (importance: 0, dev_relevance: 75)
【開催報告】フロンティア AI 時代のサイバー攻撃に備える ― 臨時開催 AWS セキュリティセミナー (importance: 35, dev_relevance: 45)
週刊生成AI with AWS – 2026/10/26 週 (importance: 40, dev_relevance: 70)
AI ガードレールによるコンタクトセンターでの責任ある AI の実装 (importance: 30, dev_relevance: 40)
AIエージェント運用の始め方！本番で止まらない管理項目・体制・改善サイクルを解説 (importance: 20, dev_relevance: 40)
AIエージェントのカスタマーサポート活用術！できることと導入事例 (importance: 15, dev_relevance: 15)
キオクシア、2桁成長に“急ブレーキ”？　Q3「売上収益1％増」、決算説明会で幹部らが語ったこと (importance: 15, dev_relevance: 10)
建設現場の「古い図面」問題　紙とデジタルのズレが招く「手戻り」と「工事遅延」、日本HPはどう解消した？ (importance: 0, dev_relevance: 15)
「保安検査の長い列」と「人手不足」をどう解消する？　フランクフルト空港におけるAI時代の空港DX (importance: 15, dev_relevance: 25)
キオクシア「設備投資を急がねば……」　2027年分は完売、メモリ供給追い付くか (importance: 30, dev_relevance: 5)
その日、メルカリは「業務を止めた」――上司と部下が対立も、AI利用率100％をかなえたCAIOの奮闘記 (importance: 35, dev_relevance: 35)
オラクルの機能がChatGPTで利用可能に　複数システムをまたぐ業務をAIで自動化 (importance: 45, dev_relevance: 40)
Metaの「Muse」とは？パーソナルAIエージェントの特徴・使い方・料金・商用利用 (importance: 40, dev_relevance: 20)
バイブハッキングとは？AIエージェントの猛威や事例・企業の対策を解説【2026年9月最新】 (importance: 25, dev_relevance: 35)
アイスマイリー、11/18（水）から3日間、「Japan DX & AX Week【関西】2026」にブース出展 (importance: 10, dev_relevance: 15)
🔒 「AIで作ればSaaSは解約できる」は本当か？作れたはいいが保守できない、を防ぐポイント (importance: 35, dev_relevance: 60)
🔒 CLAUDE.md・AGENTS.mdに書いたルールをAIに守らせるには、どの行を残し、どこへ移し、何を消すべきか (importance: 40, dev_relevance: 90)
ChatGPTのチャットを消してもライブラリやメモリに写真や個人情報が残る問題にどう対処するか (importance: 35, dev_relevance: 15)
🔒 AIエージェントの導入効果を「工数が何時間減ったか」以外の数字でも測る方法 (importance: 30, dev_relevance: 35)
CLIPは画像と文章をどう比べるのか (importance: 30, dev_relevance: 40)
「Ivanti Connect Secure」の脆弱性悪用で注意喚起 - 米当局 (importance: 0, dev_relevance: 30)
「WordPress 7.1.4」で更新時RCEや複数XSSなど脆弱性を修正 (importance: 0, dev_relevance: 40)
DB管理ツール「pgAdmin 4」に複数脆弱性 - 2件は「クリティカル」 (importance: 0, dev_relevance: 50)
QNAP、脆弱性「PHP2Shell」へ順次対応 - 「NDR」向け更新は準備中 (importance: 0, dev_relevance: 30)
個人情報漏洩の個人向け被害者相談窓口を強化 - IPA (importance: 0, dev_relevance: 5)
アイリスオーヤマ通販サイトで侵害 - 詳細を調査 (importance: 0, dev_relevance: 0)
外務省のネットワークで侵害 - 個人情報流出の可能性 (importance: 0, dev_relevance: 0)
ECサイトに不正アクセス、会員情報流出 - フィギュアメーカー (importance: 0, dev_relevance: 0)
先週注目された記事（2026年10月25日〜2026年10月31日） (importance: 0, dev_relevance: 0)