AI News Digest 2026-09-22

台本で使った記事

特集

説明に紐づくツール実行 ―― モデルの理由説明を信用せず、サーバー側でアクションを検証するAIエージェント制御
Explanation-Bound Tool Execution for AI Agents: Server-Verified Action Claims Without Trusting Model Rationales
SWE-Proof ―― 言語モデルは機械検証付きの証明で実世界の課題を解決できるか？
SWE-Proof: Can Language Models Resolve Real-World Issues with Machine-Checked Proofs?

開発者コーナー

長時間稼働AIエージェントの権限失効 ―― 委任と非同期実行下でのルート単位の権限収束
Authorization Revocation for Long-Running AI Agents: Root-Scoped Quiescence under Delegation and Asynchronous Execution

中堅コーナー

国連科学パネル「AIエージェントへの人間の支配は保証されない」と警告
UN science panel says there is "no assurance humans will keep control" over AI agents

AIツール紹介コーナー

AX ―― Googleのオープンなエージェント実行オーケストレーター
AX – Google's Open Agentic Orchestrator

速報コーナー

参考記事一覧

参考記事一覧を表示（639件）

How we made the first comprehensive map of deaths along the US border’s “virtual wall” (importance 5 / dev 0)
Trump now says he wants to form an ‘AI Force’ (importance 40 / dev 10)
Amazon blocks Meta's AI agent Muse from online shopping (importance 55 / dev 45)
UN science panel says there is "no assurance humans will keep control" over AI agents (importance 65 / dev 50)
Grok 4.7: Price, Benchmarks and How to Use It (importance 55 / dev 75) (いいね相当スコア: 0)
xAI’S Grok 4.6 Now Available In Amazon Bedrock - TradingView (importance 45 / dev 70)
Tesla to Launch Grok in the Middle East and Morocco With Upcoming Update 2026.32.6 - Not a Tesla App (importance 30 / dev 20)
Meta's Muse and xAI's Grok Bot Can Now Shop With Your Stripe Wallet - Startup Fortune (importance 50 / dev 40)
After agreeing with Anthropic CEO Dario Amodei on slowing pace of AI, Sam Altman and Elon Musk teased nex - timesofindia.indiatimes.com (importance 60 / dev 15)
DeepSeek expects training chips from Huawei in Q4 - Breakingthenews.net (importance 50 / dev 35)
AI Model Release Cycle Shrinks from 125 Days to 44 Days—Self-Improving AI Raises Control Concerns - finance.biggo.com (importance 65 / dev 60)
China, US Are Competing For AI Dominance But Have Shared Concerns Over Safety - Deccan Chronicle (importance 55 / dev 25)
Changes to LLM pricing: Baidu, Inceptron, StreamLake and Tencent (importance 0 / dev 65) (いいね相当スコア: 0)
What Sun got wrong (importance 15 / dev 35)
Attention is all you have (importance 20 / dev 25)
Turn off and restrict access to Apple Intelligence features on Mac (importance 20 / dev 30)
Python Workers are now generally available (importance 60 / dev 85)
This Digital Radio Gets Messages to the World's Remotest Locations (importance 25 / dev 30)
Kev: Tiny Jev-like family of decision models built on top of Qwen3.5 (importance 45 / dev 75)
A restored PDP-11/83 serving this page on 211BSD Unix (importance 10 / dev 20)
How do Traffic Signals Work (2019) (importance 5 / dev 10)
What happened to the Snowden archive (importance 15 / dev 10)
AX – Google's Open Agentic Orchestrator (importance 75 / dev 90)
M5 Ultra Mac Studio Review (importance 50 / dev 70)
Whirlpool Washer Transmission Repair (2007) (importance 5 / dev 10)
Help 404 Media Find Out How Your Local Police Are Surveiling You (importance 20 / dev 15)
Samsung is expected to more than double output of its HBM4 and HBM4E DRAM (importance 55 / dev 60)
Heretic removes restrictions from language models (importance 45 / dev 75)
Raspberry Pi blocks changing RAM chips (importance 30 / dev 55)
Show HN: Mini-AGI – Dynamic continual learning model trained on 8GB VRAM (importance 55 / dev 80)
Show HN: Lossless-memory – a personal AI memory that never summarizes (importance 40 / dev 75)
Advisory Group on Mathematics and Artificial Intelligence (importance 35 / dev 45)
Building standards for the next phase of AI (importance 50 / dev 40)
Expanding OpenAI Academy with new learning paths (importance 30 / dev 70)
How V7 gives AI agents institutional memory (importance 40 / dev 70)
Improving synthesis prediction of small molecules at scale with RetroChimera (importance 50 / dev 50)
Pruning LLMs Like a Physicist: Block Removal as an Ising Optimization Problem (importance 50 / dev 75)
tokenizers v1: encode, decode and scaling, measured (importance 55 / dev 80)
Open-Sourcing Rebalancer: A Generic, High-Performance Library for Solving Assignment Problems (importance 50 / dev 80)
Inside Petal: Building the World’s First Petabit-Class Transoceanic Subsea Cable (importance 45 / dev 55)
Benchmarking LLM Inference at Scale with AIPerf (importance 60 / dev 85)
Google confirms Gemini models hacked three companies in May 2026 (importance 70 / dev 65)
With Tabby, a former accountant is using AI to make accountants obsolete (importance 35 / dev 35)
Where will the next breakout startup come from? Benchmark’s full partnership weighs in at TechCrunch Disrupt 2026 (importance 15 / dev 10)
Google’s $899 Googlebook is a bet that you’ll buy a new laptop for Gemini (importance 50 / dev 60)
From first users to billions: Google’s Robby Stein joins TechCrunch Disrupt 2026 (importance 10 / dev 5)
Meet the next wave of VCs judging Startup Battlefield 200 at TechCrunch Disrupt 2026 (importance 10 / dev 5)
6 days left to save up to $200 to TechCrunch Disrupt 2026 (importance 5 / dev 5)
World model companies are keeping a lot of secrets (importance 55 / dev 40)
Is the AI industry really ready to slow down? (importance 45 / dev 20)
Vocci’s ring adds a new form factor to meeting note-taking (importance 25 / dev 25)
ScrollEd wants to turn textbooks into TikTok (importance 25 / dev 40)
Can John Ternus find Apple’s next big thing? (importance 35 / dev 20)
iPhone owners can now submit claims in Apple’s $250 million Siri AI settlement (importance 35 / dev 25)
No one is surprised that Nvidia’s Jensen Huang thinks AI fears are overblown (importance 35 / dev 15)
Podcast: Securing AI Agents: Identity, Authorization, and the DPACT Framework (importance 65 / dev 85)
Presentation: The Agent Harness: Control Planes, Invariants, and Approval Boundaries for Production AI Agents (importance 70 / dev 90)
Alibaba Open Sources OpenCodeReview for AI-Assisted Code Review (importance 60 / dev 85)
Google Agent Development Kit for Kotlin Reaches Feature Parity with Python, Supports On-Device AI (importance 65 / dev 90)
ByteDance launches Dramagic, a full-pipeline AI platform for producing short dramas from script to screen (importance 45 / dev 50)
SoftBank to borrow over $11 billion in risky bonds for OpenAI stake (importance 50 / dev 15)
Bristol researchers say medicine already knows how to handle black boxes and AI could learn from it (importance 55 / dev 55)
US and China agree on AI dialogue with security mechanism ahead of Trump-Xi summit (importance 55 / dev 30)
Alibaba's open-weight Qwen-Image-2.1 claims to beat closed models in image generation with just 7 billion parameters (importance 0 / dev 0)
Tencent's Gander aims to keep talking while it works in the background (importance 0 / dev 0)
Runway wants to turn AI video generation into a live stream you control in real time (importance 0 / dev 0)
RBS-Attention: Radius-Bounded Sparse Prefill for Long-Context Large Language Models (importance 50 / dev 80)
Attention-Aware Routing: Coupling Routing and Attention in MoEs (importance 50 / dev 80)
CaLR: Causal Latent Revision for Robust Diffusion Reasoning (importance 45 / dev 75)
LoRA Enhanced Contrastive Learning with SAS Vision Transformers (importance 35 / dev 70)
Detecting Hallucination in LLMs: Tracing the Topological Signatures of Impaired Context Sharing (importance 55 / dev 80)
Decoupling Internal Representational Changes and Causal Importance in Fine-Tuned Large Language Models (importance 50 / dev 80)
TinyCeNN-LM: Quality-Gated Conversion of Pretrained Attention with CeNN-Inspired Cellular-Recurrent Layers (importance 45 / dev 75)
Clinician-Grounded Quality Assurance for AI-Assisted Psychiatric Intake (importance 45 / dev 60)
Can Agents Design Better Chips with a Higher Level Abstraction? (importance 55 / dev 85)
SpecOpt: Contact-Diff Reasoning for Agentic Molecule Optimization Toward Binding Specificity (importance 45 / dev 60)
Implicit Rule Induction with Test-Time Task Embeddings in ARC-like Tasks (importance 40 / dev 70)
AI-GRACE: A Use-Case Operationalization Framework for Agentic AI: From Organizational Objectives and Obligations to Deployment Capabilities and Architecture (importance 60 / dev 85)
Information-Gain Rewards over Diversity-Pruned Tests: GT-Anchored Verifier Co-Training for Reliable Code Generation (importance 50 / dev 85)
Ability-Residual Decoupled Modeling for Affective Cognitive Diagnosis (importance 35 / dev 60)
A Fully Differentiable Neuro-Soft-Symbolic Framework for Perceptual Task Planning (importance 45 / dev 80)
CogGym: Towards Large-Scale Comparative Evaluation of Human and Machine Cognition (importance 50 / dev 75)
PlaceReasoner-Beta: Reasoning-Driven Macro Placement and Benchmarking (importance 40 / dev 80)
Efficient Benchmarking in Production: A Study of an Evolving LLM Agent (importance 55 / dev 85)
GameASG-Bench: Benchmarking Autonomous Software Generation for Game Development (importance 45 / dev 80)
LEGIT: Credentialing Protocol for Trustworthy AI Agent Marketplaces (importance 55 / dev 80)
Offline Multimodal Large Language Models for Decision Support in Air Operations (importance 50 / dev 75)
DENSE: Distilling Agent Trajectories into Evidence-Grounded Shortcut Trees for Self-Refinement (importance 55 / dev 85)
GVPO++: Group Variance Policy Optimization for LLM Post-Training and On-Policy Distillation (importance 55 / dev 85)
Risk-Aware Occupancy for Safety-Oriented End-to-End Autonomous Driving (importance 45 / dev 75)
Driving on Registers, Reasoning on Risk: Risk-Aware Occupancy for Register-Based End-to-End Autonomous Driving (importance 40 / dev 75)
LogicTrack: Auditing Reasoning Trajectories of Large Language Models with Formal Logic Solvers (importance 55 / dev 80)
PolyBridgeBench: Benchmarking Multimodal LLMs for Physics-Grounded Bridge Design (importance 45 / dev 75)
The Communication Bottleneck: A Round-Trip Study of Tree-Structured Expression Serialization in Language Models (importance 50 / dev 80)
Learning-to-Optimize as the Missing Architectural Layer of AI-Native Networks (importance 50 / dev 85)
Dual-Interest Sequential Product Recommendation With Multi-Granular SSM (importance 40 / dev 75)
Beyond Accuracy: Centroid-Guided Contrastive Loss for Structured Fraudulent Job Posting Detection (importance 35 / dev 70)
Reducing Barriers to Academic Support: Evaluating a Course-Specific RAG System for Addressing Help-Seeking Disparities in Higher Education (importance 45 / dev 75)
Calibrating Teacher--Student Discrepancy for On-Policy Distillation (importance 50 / dev 85)
One Prompt Does Not Fit All: Self-Meta-Evolve for Personalized Information Extraction (importance 50 / dev 80)
Accelerating Dense LLMs via L0-regularized Mixture-of-Experts (importance 55 / dev 80)
GUARD: Natural Forgetting in Large Reasoning Models via Guided Answer-Reasoning Distillation (importance 50 / dev 70)
Listen Before You Speak: Response Planning from Listener Facial Reactions for Conversational Speech Generation (importance 35 / dev 60)
World Modeling in Transformers (importance 60 / dev 75)
ECG Mirage: Revealing and Mitigating the Underutilisation of ECGs in Vision-Language Models for Clinical Prediction (importance 40 / dev 50)
LLM-Generated Feature Pools for Time Series Anomaly Detection (importance 45 / dev 70)
MIST: Multimodal Survival Prediction with Genomic-Guided Histology Attention (importance 35 / dev 50)
EnterpriseVal: Quantifying the Efficacy, Reliability and Value of Generative AI in the Enterprise (importance 65 / dev 65)
AutoRecLab: Describe the Experiment, Get the Code! (importance 55 / dev 75)
What Should We Ask Next? Retrieval-Aware Question Learning under Partial Evidence (importance 45 / dev 70)
AutoViewMem: Self-Configuring Orthogonal Views for Conversational Long-Term Memory (importance 65 / dev 75)
Learning Cardiac Features: ECG Biometrics Across Time and~Exercise (importance 35 / dev 45)
A Lie Detector Test for Language Models: Reading Knowledge a Model Won't Reveal (importance 70 / dev 80)
CodeMidas: Scaling Agentic Coding RL Environments from Code Itself (importance 75 / dev 85)
Designer-RSI: Evolving Procedural Memory from User Traffic for Agentic Graphic Design (importance 60 / dev 75)
ResNLS: An Improved Model for Stock Price Forecasting (importance 25 / dev 30)
dSTAR: Straggler Tolerant and Byzantine Resilient Distributed SGD (importance 55 / dev 75)
A Hybrid Computational Intelligence Framework for scRNA-seq Imputation: Integrating scRecover and Random Forests (importance 35 / dev 40)
Making Latent Evolution Explicit: Operator-Structured Transitions for World Action Models (importance 55 / dev 75)
Reinforcement learning for post-coronagraphic wavefront control (importance 40 / dev 50)
BI-Agent and BI-Bench: Towards Automating End-to-End Business Intelligence (importance 70 / dev 75)
SpaceDiffusion: Over-the-Orbit Diffusion for Space Generate-and-Forward Communications (importance 40 / dev 50)
Bio-MF: Low-Latency and High-Fidelity EEG-to-fNIRS Cross-Modal Generation for Hybrid Motor-Imagery Brain--Computer Interfaces (importance 35 / dev 40)
Trustworthy FinAInce: Unpacking How AI-Mediated Financial Advice is Judged (importance 50 / dev 40)
Scaling Discovery through Test-Time Communication (importance 70 / dev 80)
Physically Based Rendering in the Latent Space (importance 50 / dev 70)
How Much of a Real Workload Can LLM-Generated GPU Kernels Actually Reach? (importance 75 / dev 90)
PlantShade: Predicting Plant Shadows for Lighting-Aware Robotic Agricultural Operation (importance 45 / dev 60)
Aligning with Lived Experience: Heterogeneous Benefits of Fine Tuning in Mental Health Support Generation (importance 55 / dev 60)
Geometry of Values: Task Vector Composition for Ethical Preference Alignment in Language Models (importance 70 / dev 80)
From Task Success to Productive Success: Evaluating Human-AI Collaboration by Quality and Cost (importance 65 / dev 70)
The Stochastic Shift: A New Evaluation Paradigm for Text-to-SQL with AI Operators (importance 55 / dev 75)
EnSol: an environment-aware graph neural network for molecular solubility prediction (importance 40 / dev 55)
SWE-Proof: Can Language Models Resolve Real-World Issues with Machine-Checked Proofs? (importance 80 / dev 85)
Visual Navigation Transformer with Pose Attention (importance 50 / dev 70)
Fewer Steps, Better Actions: Rethinking Flow-Matching Inference for VLA Policies (importance 55 / dev 75)
Hallucination-R1: Robustness-Oriented Paraphrase Generation for Factual Consistency (importance 65 / dev 75)
FOCAL-VLA: Subtask-Guided Geometry Distillation and Implicit World Modeling for Vision-Language-Action Models (importance 65 / dev 80)
KnowDemo: Knowledge-Guided Robot Demonstration Generation from Human Videos (importance 60 / dev 75)
VLA-Scope: Shift-Aware Failure Prediction for Vision-Language-Action Models (importance 70 / dev 80)
Verify, Don't Trust: Agentic Model Development for Video Discovery Retrieval at Scale (importance 75 / dev 85)
Beyond Exact Match: Task-Aware GRPO for Cross-Domain PCBA Visual Question Answering (importance 50 / dev 70)
Authorization Revocation for Long-Running AI Agents: Root-Scoped Quiescence under Delegation and Asynchronous Execution (importance 80 / dev 85)
Deep Reinforcement Learning with Buffered Quantile Objectives (importance 50 / dev 75)
Co-Evolving Zero-Day Jamming: Adaptive Attack Synthesis and Graph Attention-Based Online Detection (importance 50 / dev 70)
CESBench: Benchmarking Large Language Models on Cryptographic Engineering Security for IoT Devices (importance 70 / dev 85)
From Memory to Behavior: A Behavior-Aware Role-Playing Framework for Social Media Influencers (importance 40 / dev 50)
Knowledge-Graph-Augmented Chronos-2 for HEC-RAS Surrogate Forecasting (importance 50 / dev 65)
AgentVidBench: A Multi-Hop Video Question Answering Benchmark for Evaluating MLLM Agents (importance 70 / dev 80)
Consistent Relexicalization of Clinical Documents using Graph-Based Approach (importance 55 / dev 70)
WS-NeRF: A Mamba-Driven World-State-Aware Adaptive Deblurring Neural Radiance Field (importance 50 / dev 75)
Talking Past the Machine: Morality, Politeness, and Alignment in Human-AI Dialogue (importance 60 / dev 70)
Think Locally, Refine Globally for Memory-Efficient 3D Reconstruction (importance 55 / dev 75)
Interference-Driven Clustered Optimisation for FM Spectrum Coordination (importance 45 / dev 60)
AtomEgo: Exploring Ego-Robot Integration for Embodied Foundation Model Pretraining (importance 75 / dev 85)
OmniVChat: Synthesizing, Benchmarking, and Training for Native Audio-Visual Dialogue (importance 70 / dev 80)
HE-Guardrail: A Homomorphic Guardrail Against Jailbreak Attacks for Encrypted Large Language Model Inference (importance 80 / dev 85)
2nd Place Solution to the HANDS 2026 Workshop Challenge-Dexterous Grasp Motion Track: Single-Shot Trajectory Warping for Grasp Motion Generation (importance 45 / dev 70)
VidOmni-Bench: A Benchmark for Fine-Grained Video Understanding via Spatio-Temporal Event Verification across Complexity and Duration (importance 70 / dev 75)
OneBid: A Unified Auto-Bidding Foundation Model for Diverse oCPX Advertising Scenarios (importance 60 / dev 70)
On Repulsive and Attractive Teachers: Separating Correctness from Behavior in Self-Distillation (importance 65 / dev 80)
GameLogicBench: Evaluating Coding Agents on Runtime Game Logic with Tick-Level State Assertions (importance 70 / dev 85)
CityLearn v3: A Configurable Simulation and Evaluation Framework for Realistic Control Studies of Renewable Energy Communities (importance 50 / dev 70)
Micro-Collaborative Poisoning: A Distributed Attack on RAG Systems (importance 75 / dev 85)
Potential-Field Action Representation for Reinforcement Learning in Contact-Rich Manipulation (importance 60 / dev 80)
Steering LLMs Responses Towards Moral Foundations on the Norwegian MFQ-30 (importance 50 / dev 65)
Chinese Competitive Debating Dataset and Benchmark (importance 50 / dev 70)
SynthDemo-RL: Breaking the Zero-Reward Barrier in VLA Adaptation with LLM-Guided Synthetic Demonstrations (importance 70 / dev 85)
Outcome-Conditioned End-Effector Geometry Across Vision-Language-Action Policies (importance 55 / dev 75)
When Steering Fails in Latent Reasoning: A Latent-to-Language Transition Gap (importance 70 / dev 80)
Samsone: A Family of Open Small Audio Language Models for On-Device Inference (importance 75 / dev 85)
From Code Archival to Knowledge Graph: Bridging Software Heritage, COAR Notify and Wikidata (importance 50 / dev 65)
CIPL: A Channel-Aware Framework for Recoverable Privacy Leakage in LLM Agents (importance 75 / dev 85)
TERMon: Detecting Persistent Behavioral Threats in Edge AI via Hardware-Native Ternary Runtime Monitor (importance 75 / dev 85)
CIBuzzBench: A Benchmark for Cross-Lingual Understanding of Chinese Internet Buzzwords (importance 50 / dev 70)
Balanced Prompt Adaptation against Entropy-Induced Collapse for Test-Time Binary Segmentation (importance 55 / dev 75)
ForceTwin: Physics-informed Digital Twins for Robotic Manipulation from Instrumented Human Interaction (importance 70 / dev 85)
An Agentic Just-in-Time Adaptive Intervention System for Personalized Sleep Support: Proof-of-Concept Study with N of 1 Data (importance 50 / dev 65)
Matrix AdaGrad: Row-wise and Column-wise Adaptive Subgradient Methods (importance 55 / dev 80)
Touvigation: Embodied Adaptive Object Acquisition for Blind and Low-Vision Users in Unfamiliar Indoor Environments (importance 55 / dev 75)
Federated Deep Clustering Networks for High-Dimensional and Heterogeneous Data (importance 55 / dev 80)
Do Personality-Tuned LLMs Make Better Social Agents? (importance 55 / dev 70)
Neural Cellular Automata Learn General Features in their Hidden Channels (importance 55 / dev 75)
Benchmarking the Explanatory Quality of Open-Weight Vision-Language Models in Face Recognition (importance 60 / dev 75)
Detecting Pretraining Data in Large Language Models from a Free-Energy Perspective (importance 75 / dev 85)
When Should a Failing Robot Ask? Initiating Corrective Human-Robot Dialogue from Audited Sensor Evidence (importance 65 / dev 80)
NemotronLabs VoiceChat: An Open Full-duplex Speech-to-Speech Model with Tool Calling Capabilities (importance 80 / dev 90)
Bayesian Belief Layer for Controllable Opinion Dynamics in LLM Agents (importance 70 / dev 85)
DiaVLo: Diagnosing Behaviours of Vision-Language Models (importance 70 / dev 85)
Gricea: An Open Science Platform for Conversational AI Research (importance 65 / dev 80)
Value-Sensitive Delegation in Everyday AI Agent Use: Evidence from OpenClaw (importance 65 / dev 75)
Collab-Solver: Collaborative Solving Policy Learning for Mixed-Integer Linear Programming (importance 60 / dev 75)
Fact Grounded Attention: Eliminating Hallucination in Large Language Models Through Attention Level Knowledge Integration (importance 75 / dev 85)
MemeLens: Multilingual Multitask VLMs for Memes (importance 55 / dev 75)
Beyond Final Answers: CRYSTAL Benchmark for Transparent Multimodal Reasoning Evaluation (importance 70 / dev 80)
Transferable knowledge graphs with executable learned operators for algorithm design (importance 65 / dev 85)
BoostAPR: Boosting Automated Program Repair via Execution-Grounded Reinforcement Learning with Dual Reward Models (importance 75 / dev 90)
On the Limitations of Large Language Models for Conceptual Database Modeling (importance 60 / dev 75)
Intent-Governed Tool Authorization for AI Agents (importance 75 / dev 85)
Explanation-Bound Tool Execution for AI Agents: Server-Verified Action Claims Without Trusting Model Rationales (importance 80 / dev 90)
A Forced-Structure Reduction and Verifiable Bounds for Conway's 99-Graph (importance 40 / dev 5)
Planetary Prediction Engine: Autonomous Geospatial Prediction via Intelligent Data Selection and Foundation Model Embeddings (importance 50 / dev 40)
Balance of Benchmarks: Semantic Density Reweighting for Task-Conditioned Model Comparison (importance 35 / dev 60)
A visual large language foundational model for medical image recognition using clinician-contributed online resources (importance 50 / dev 40)
Fraglingo: Molecular Design via Attachment-Aware Autoregressive Fragment Generation (importance 45 / dev 30)
MOSCOPT: Mixture-of-Skills Collective Optimization for LLM Agents (importance 55 / dev 70)
Runtime Authorization for Resources Acquired by AI Agents (importance 60 / dev 75)
Collaborative Memory for Multi-Agent VLM Systems (importance 55 / dev 70)
Bad Genius: Counterfactual-Guided Harness Evolution Beyond Task-Specific Shortcuts (importance 55 / dev 75)
Disentangling Long-Term Memory via Latent Neuro-Symbolic Reasoning (importance 50 / dev 70)
A Unified Evaluation Framework for Trustworthy Large Language Models, Agentic AI, and Multimodal Systems (importance 55 / dev 75)
Rethinking Multi-Agent Collaboration: When More Is Less (importance 55 / dev 75)
NeuSOGA3D: A Neuro-Symbolic Framework for Explainable 3D Geometric Reconstruction (importance 45 / dev 60)
Reinforcement Learning under External Influence: Guarantees, Algorithms, and Sample Complexity (importance 40 / dev 70)
Soda: An Object-Oriented Functional Language for Specifying Human-Centered Problems (importance 35 / dev 75)
Continuous Spiking Graph Neural Networks (importance 40 / dev 65)
Understanding In-context Learning of Addition via Activation Subspaces (importance 50 / dev 70)
AntiGrounding: Executable Robot Trajectories as Visual Prompts for VLM-Guided Manipulation (importance 45 / dev 65)
Generalizing Beyond Suboptimality: Offline Reinforcement Learning Learns Effective Scheduling through Random Solutions (importance 40 / dev 65)
Benchmarking Autonomous Driving Planners Across Leaderboards: A Unified CARLA-Based Evaluation (importance 45 / dev 65)
Auditing a KB Elicitation of Frontier LLM Knowledge: A Multi-dimensional Analysis of GPTKB v1.5 (importance 50 / dev 70)
The Impact of Semantic Pairs on Self-Supervised Representation Learning (importance 40 / dev 65)
Deep Learning-Enhanced Real-Time Wi-Fi Sensing Through Single Transceiver Pair (importance 35 / dev 60)
Understanding Structural Representation in Foundation Models for Polymers (importance 45 / dev 50)
Large Language Models As Shannon Lossy Compressors Not Solomonoff Induction Estimators: The Singularity Is Not Near Without Symbolic Model Synthesis (importance 55 / dev 75)
BEAT-Net: Injecting Biomimetic Spatio-Temporal Priors for Interpretable ECG Diagnosis (importance 40 / dev 50)
HERMES: A Holistic End-to-End Risk-Aware Multimodal Embodied System with Vision-Language Models for Long-Tail Autonomous Driving (importance 45 / dev 60)
MENASpeechBank: A Reference Voice Bank with Persona-Conditioned Multi-Turn Conversations for AudioLLMs (importance 50 / dev 70)
Position: A Dynamical Systems Perspective is Needed to Advance Time Series Modeling (importance 40 / dev 70)
The MAMA-MIA Challenge: Advancing Generalizability and Fairness in Breast MRI Tumor Segmentation and Treatment Response Prediction (importance 40 / dev 50)
Taming the Adversary: A Cost-to-Disturbance Ratio Approach to Adversarial Reinforcement Learning (importance 40 / dev 70)
Towards the Vision-Sound-Language-Action Paradigm: The HEAR Framework for Sound-Centric Manipulation (importance 50 / dev 70)
How do LLMs Compute Verbal Confidence (importance 50 / dev 75)
Evolving Skill Modules under a Fixed Planner: Versioning, Rollback, and Runtime Governance for Long-Lived Robot Systems (importance 50 / dev 75)
Lessons Without Borders? Evaluating Cultural Alignment of LLMs Using Multilingual Story Moral Generation (importance 40 / dev 50)
Representation Before Training: A Practical Benchmark for Generative Medical Event Model Tokenization (importance 40 / dev 60)
Diagnostic-Guided Longitudinal Modeling for Forecasting Retinal Atrophy Progression (importance 40 / dev 60)
How a Cooperative-Override Circuit Suppresses Nash Play in Large Language Models (importance 50 / dev 75)
Why Do LLMs Struggle in Strategic Play? Broken Links Between Observations, Beliefs, and Actions (importance 50 / dev 75)
REALM: An RGB- and Event-Aligned Latent Manifold for Cross-Modal Perception (importance 40 / dev 65)
Rhamba: Region-Aware Hybrid Attention-Mamba Framework for Self-Supervised Learning in Resting-State fMRI (importance 40 / dev 60)
Constraint Decay: The Fragility of LLM Agents in Backend Code Generation (importance 55 / dev 80)
LiteMedCoT-VL: Parameter-Efficient Adaptation for Medical Visual Question Answering (importance 40 / dev 50)
The critical slowing down in training diffusion models (importance 50 / dev 75)
Revisiting Reinforcement Learning with Verifiable Rewards from a Contrastive Perspective (importance 55 / dev 80)
PaCo-VLA: Passivity-Shielded Compliance Prior for Contact-Rich Vision-Language-Action Manipulation (importance 50 / dev 75)
AgenticRL: Agentic Reinforcement Learning with Self-Refinement for Complex UAV Navigation (importance 45 / dev 70)
Scaling Novel Graph Generation via Lightweight Structure-Guided Autoregressive Models (importance 45 / dev 70)
WorldRoamBench: An Open-World Benchmark for Long-Horizon Stability of Interactive World Models (importance 50 / dev 75)
Learning Gait-Aware Quadruped Locomotion with Temporal Logic Specifications (importance 40 / dev 70)
GeoSelect: Spatial-Program Execution for Training-Free Referring Remote Sensing Image Segmentation (importance 40 / dev 65)
Self-Reference in Large Language Models: The Introspection Threshold for Recursive Self-Improvement (importance 55 / dev 80)
Prompt-Driven Exploration: Language as an Exploration Space for VLA Reinforcement Learning (importance 50 / dev 75)
Git-Assistant: Planning-Based Support for Updating Git Repositories (importance 55 / dev 85)
Cover First, Disagree Softly: Rethinking Mismatch-First Active Learning for Frame-Level Audio Classification (importance 35 / dev 65)
Teaching LLMs to Self-Evolve: Cultivating Core Meta-Skills with Reinforcement Learning (importance 55 / dev 80)
Sixteen models, fewer than two voices: measuring ensemble dispersion where no answer is uniquely correct (importance 40 / dev 70)
Are You Sure You're Sure? On the Impact of Instruction Tuning on Confidence and Lexical Diversity (importance 45 / dev 70)
Multi-turn Conversational AI from Text to Multimodal Interaction: Data, Models, Evaluation, and Open Challenges (importance 55 / dev 75)
Modeling Human Behavior with Type Vectors Using AI (importance 40 / dev 60)
Self-Explanation Tutor for Active Study of CS1 Worked Examples (importance 40 / dev 60)
A Training-Free Proactive Defense Against Partial Speech Manipulation via Self-Embedding Steganography (importance 35 / dev 65)
PAVE: Predictive Alignment and Value-Guided Evolution for World-Action Policies (importance 45 / dev 75)
Position Matters: Feature Inversion Attacks in ViT Split Inference with Token Reduction and Shuffling (importance 40 / dev 70)
TabScope: Question-Adaptive Scope Selection for Table Question Answering (importance 45 / dev 70)
VLA-Precision: Asymmetric Co-Bootstrapping for Efficient Real-World Online RL of Vision-Language-Action Models (importance 50 / dev 75)
Staying on the Attack Path: Structured State for Long-Horizon Automated Penetration Testing (importance 50 / dev 80)
Fine PT-PT Web: A High-Quality 41 Billion Tokens Data Collection of the European Portuguese Web (importance 35 / dev 65)
Do New Attention Mechanisms Actually Fix Attention Sinks at Million-Token Context? (importance 55 / dev 80)
High-probability guarantees for linear accessibility in feature superposition (importance 40 / dev 70)
Data-free On-policy Distillation (importance 55 / dev 80)
A primer on evaluation methods for large language models in healthcare (importance 45 / dev 65)
From Momentary Emotion Inference to Sustained Emotion Support: Evaluating a Companion Agent in a Longitudinal Study (importance 45 / dev 60)
PentestChain: A Cost-Aware, MCP-Orchestrated Framework for Automated Penetration Testing with Free-Tier LLMs (importance 50 / dev 80)
CPR: Combining global composing, local performing and full-sequence refining in piano rendering with continuous autoregressive modelling (importance 35 / dev 50)
Knowledge-Graph Based Augmentation versus Retrieval Augmented Generation for Cultural-Related Question Answering (importance 45 / dev 70)
ASLEval: Measuring Privacy Exposure Displacement in LLM Agent Sessions (importance 50 / dev 80)
Large Language Model Agents for Evidence Based Genetic Disease Severity Classification (importance 40 / dev 60)
CoReLoop: Parameter-Efficient Controlled Recurrent Refinement for Audio Deepfake Detection (importance 35 / dev 65)
PACE: Precise AI Cinematic Expression (importance 35 / dev 40)
Sparse Priors for Efficient Distribution Learning (importance 50 / dev 75)
Elastic Threshold Attention: Learned Contextual Sparsity for Long-Context Decoding (importance 60 / dev 80)
Continuous Delayed-Memory Stochastic Gradient Descent and Continuous-Time Reinforcement Learning from History of Astrophysical Time Series Studies (importance 35 / dev 50)
Do Quantum Models Scale Like LLMs? (importance 50 / dev 75)
When AI Reviews Train AI Reviewers: Scientific-Judgment Collapse and Mitigation (importance 50 / dev 70)
Efficient Bayes-Adaptive Reinforcement Learning with Temporal Logic Specifications (importance 40 / dev 70)
From Switching to Dynamic Regret: A Simple Reduction via Unbiased Random Sequences (importance 35 / dev 70)
Generative inversion for early ranking of competing geologic interpretations (importance 35 / dev 50)
ASGARD: Action-Space Guard for UAV Resilience via Reinforcement Learning (importance 40 / dev 70)
From Stress to Affect: Multimodal Deep Learning for Physiological Emotion Recognition Across Wearable Sensor Modalities (importance 35 / dev 60)
MOSAIC-SR: Transformer-Guided Symbolic Regression for Scientific Equation Recovery (importance 50 / dev 75)
On the Limits of Maximal Coding Rate Reduction for Out-of-Distribution Generalisation (importance 40 / dev 75)
Stiefel-AdamW: Geometry-Aware AdamW for Linear Factorization Blocks (importance 50 / dev 80)
A Lightweight Plug-in Gate for Transformer-Based Time-Series Forecasters (importance 45 / dev 75)
FedeRage: Provably Convergent Agnostic Federated Learning under General Client Drift (importance 40 / dev 75)
Toward individual-level calibration in affect recognition with perceptual adjustment queries (importance 35 / dev 60)
REFINEPPO: Learning Continuous Control Policies by Iterative Action Refinement (importance 45 / dev 75)
Talk to Me, Jarvis: An Open-Source Edge-Deployable Voice Assistant Framework for Autonomous Racecars (importance 40 / dev 70)
Signal-Centric Remote Sensing via Alternative Preprocessing and Acoustic Processing for ML-Driven Applications (importance 30 / dev 50)
Layerwise Decoupling for Stable Structured Sparsification of Fully Connected Layers (importance 45 / dev 75)
HMB-GAN: Hybrid Multi-B\'ezier GAN for Vector Shape Synthesis (importance 15 / dev 25)
M2G-LLM: Enhancing Clinical Prediction via Multimodal Graph Reasoning and LLM Context Injection (importance 20 / dev 30)
TierKV: Long-Context On-Device LLMs via Predictive Multi-Tier KV Caching (importance 70 / dev 75)
Reliability-Centered Evaluation of Sparse Longitudinal CT Lesion-Size Forecasting with Conformal Interval Calibration and Gompertz-Inspired Regularization (importance 20 / dev 20)
MIRCID: Inferred Hub-miRNAs Drive Cross-Task Improvements in Drug Mechanistic Modeling (importance 15 / dev 20)
Multi-Subject Pretraining Enables Short-Calibration Personalization for Closed-Corpus Surface EMG Speech Decoding (importance 20 / dev 25)
FairLMs: A Turnkey Library for Fairness in Language Models (importance 55 / dev 80)
Fast And Accurate Text Content File Type Identification (importance 30 / dev 40)
An Introduction to Compression-Based Machine Learning (importance 35 / dev 70)
Routine Blood Tests Outperform CRP for Distinguishing Bacterial From Viral Infection in Children (importance 20 / dev 10)
IntBMoE: Integrating Block-Level Conditioning into Expert Composition for Full-Participation Mixture-of-Experts (importance 55 / dev 70)
Probabilistic Forecasting of Business Process Executions with Neural Temporal Point Processes (importance 35 / dev 60)
Tracing the Evidence Behind Zero-Shot Time-Series Forecasting: A Source-First Taxonomy and Audit Framework (importance 40 / dev 70)
Decision-Focused Learning for Mean-Variance Portfolio Optimization via KKT-Based Reformulation (importance 30 / dev 60)
Optimal Randomized Proper Online Learning (importance 25 / dev 75)
Understanding LLM Quantization through Activation-Guided Compensation and Orthogonal Residuals (importance 60 / dev 80)
Efficient Architecture Search under Leave-One-Subject-Out Evaluation (importance 30 / dev 70)
What Must Survive? Exact Task-Information--State Frontiers for Resource-Sufficient Learning (importance 20 / dev 75)
IncentRL: The Trade-Off Between Preference Guidance and Task Performance (importance 35 / dev 75)
OpenMAS-GCom. A Diagnostic Benchmark for Graph-enhanced Multi-Agent Systems (importance 60 / dev 80)
MACE: Memory-Agent Co-Evolution with Adaptive Memory Graphs for Multi-Agent Systems (importance 60 / dev 80)
Trading Depth for Time in Recurrent Transformers (importance 45 / dev 75)
Riemannian Neural Hamiltonian Flows: Geodesic Symplectic Transport and Interpretability (importance 30 / dev 70)
Beyond Gaussian Worlds: Latent Geometry Matters for JEPAs (importance 35 / dev 75)
Multi-Domain Clustering via Measure Quantization (importance 25 / dev 60)
Optimization Geometry of Equivalent Brownian RKHS Representations (importance 20 / dev 70)
SpecQuant: Speculative Decoding with Multi-Parent Quantization for Adaptive LLM Inference (importance 65 / dev 80)
GEM-MPC: Balancing Exploration and Exploitation through Expert-Guided Planning (importance 40 / dev 75)
GraphSkillEvo: Evolutionary Optimization of Graph-Structured Agent Skills (importance 60 / dev 80)
Bilevel Optimization of Topology and Hyperparameters (BOTH) (importance 35 / dev 70)
RegKT: Interpretable and Robust Deep Knowledge Tracing With IRT-Regularizer (importance 25 / dev 70)
The Weight Is Over - Interactive Diffusion on Consumer GPUs (importance 50 / dev 75)
Geometric Mean Pooling for Equal-Weight Multiplicative Coarse-Graining (importance 20 / dev 70)
LLMs as Feature Engineers for Text-and-Tabular Prediction (importance 45 / dev 75)
ExpBoN: Exponential-Noise Best-of-$n$ for Efficient Test-Time LLM Alignment (importance 50 / dev 75)
Intervention Granularity Matters: Coherent Treatment Bundles in Counterfactual Simulation with Clinical World Models (importance 25 / dev 70)
Beyond Kinematics: Benchmarking Simulation Fidelity for Muscle-Driven Imitation Learning (importance 30 / dev 70)
Kinks vs. Smoothness: Identifiability of Real Analytic nICA for Laplace-like Sources (importance 25 / dev 75)
Joint Remaining Useful Life Prediction and Capacity Estimation of Lithium-Ion Batteries Using Partial-Charging Data (importance 20 / dev 60)
Learning to Move Cities: Deep Meta-Models and Reinforcement Policies for Calibration and Control in Urban Networks (importance 25 / dev 70)
RACER: Role-Aligned Competence Estimation for Human-AI Routing (importance 45 / dev 75)
Time series generation with spectrally aligned latent flow matching (importance 35 / dev 70)
Assessment of Machine Learning-Based Critical Heat Flux Models in the CTF Subchannel Code for Square Rod Bundle Prediction (importance 20 / dev 60)
Abstention and Noise Filtering: Two Missing Primitives of Softmax Attention (importance 50 / dev 80)
COMPLEX: A Closed-Form Certified Embedding of Multiparameter Persistence Modules (importance 20 / dev 75)
$\lambda$-Controlled GRPO: Turning Flow-Matching Ratio Instability into a Budgeted Resource (importance 50 / dev 75)
Available Guardrails: Certifying Selective Prediction across ML Systems (importance 45 / dev 75)
Particle Competition and Cooperation for Robust Graph Convolutional Network Learning Under Label Noise (importance 25 / dev 70)
Benchmarking World Models for Continual Learning on Compositional Tasks (importance 45 / dev 75)
BrainWideBench: Benchmarking large-scale pretraining and across-animal transfer in multi-region neural recordings (importance 30 / dev 75)
Training-Adaptive Convolutional Sparse Coding via Information Bottleneck for Robust Visual Representation (importance 30 / dev 70)
HERMES: Contrast-Aware Knowledge Graph Reasoning from Clinical Notes for Patient Outcome Prediction (importance 30 / dev 70)
TALON: A Temporally Aware Longitudinal Framework for Radiology Report Generation (importance 25 / dev 70)
Beyond WER: Entity and Disfluency Recall in Accented Conversational ASR (importance 40 / dev 75)
Recursive Language Models Generalize Out of Domain (importance 45 / dev 80)
PhysioBench: A Unified Benchmark for Physiological Signal Question Answering (importance 25 / dev 75)
VISPATH: Visual-Intent-Guided Path Reasoning for Multimodal Knowledge Graph Question Answering (importance 35 / dev 75)
Reading Less While Writing: A Closed-Form Bandwidth Dial for Streaming Multimodal Decoders (importance 45 / dev 80)
Rewarding Efficient Reasoning Improves Abstention on Underspecified Tasks in Reasoning Models (importance 45 / dev 80)
Reconstruction of 4D Mitral Regurgitation Hemodynamics from Sparse Planar Data using Deep Operator Networks with Test-Time Adaptation (importance 20 / dev 70)
Automated Physics-Informed Neural-Networks-Based Calibration of Highly Segmented Silicon Telescopes (importance 20 / dev 70)
The Refutation Gap: Certifying Both Halves of an Optimality Claim (importance 40 / dev 80)
Extreme classification: beating chance with one training example from each class (importance 25 / dev 75)
Complex Problem Solving in Large Language Models: A Statistical Control Survey and Diagnostic Framework (importance 50 / dev 80)
Aggregated Posterior Predictive Checks for Generative Modeling (importance 35 / dev 75)
Fragment-Aware Vision Transformers for Fresco-Fragment Style Classification (importance 15 / dev 70)
A Smoothed Discrepancy Principle for Random Feature Methods and Neural Networks (importance 25 / dev 75)
Triply-Scalable Equivariant Gaussian Process Modeling (importance 35 / dev 75)
Diverse and Adaptable Arm Coordination for Octopus-Crawling via Diffusion-Based Uncertainty-Aware Optimization (importance 20 / dev 70)
Multiclass Semantic Segmentation of Wildland Fire Images Using Context-Aware Centralized Copy-Paste Data Augmentation (importance 20 / dev 70)
From Trainability Diagnostics to Optimization Claims: Boundaries and Controls in Variational Quantum Optimization (importance 25 / dev 75)
Programming AMD XDNA NPUs with Open-source Compiler Tools: A FlashAttention Case Study (importance 60 / dev 85)
How Many Humans Is a Judge Panel Worth? (importance 50 / dev 75)
Hybrid GPU-CPU Retrieval for Personalized Search at Ultra-Large Scale (importance 60 / dev 80)
Identifying Security Platform Product Abuse with Machine Learning (importance 40 / dev 70)
Diagonalized Attention for Individualized Regression: Latent-Row Localization and Prediction (importance 30 / dev 75)
Sparse Identification for Automatic Large-Scale Screening: A Constraint-Aware Framework with Ultra Fast Decoding Algorithm (importance 25 / dev 70)
Hiding in Plain Sight: A Diffusion-based Mitigation of Geolocation Privacy Leakage in Vision-Language Models (importance 50 / dev 75)
Prediction Dynamics in Depth-Recurrent Language Models (importance 50 / dev 80)
Brownian Heads for Deep ReLU Representations: Activation Mass and the Cost of Same-Sample Selection (importance 25 / dev 75)
FootQuery: Future-Touchdown-Guided Retrieval from Depth History for Perceptive Humanoid Locomotion (importance 20 / dev 70)
Improving the Predictive Performance of Bootstrap Aggregating by Dirichlet Resampling (importance 30 / dev 75)
Adaptive Rollout Truncation Based on Epistemic Uncertainty for Efficient Offline World Model Training (importance 40 / dev 75)
ServeGuard: Verifiable, Bounded-Residual Confinement of Operator-Invisible Channels Without Revealing the Certified Read Factor (importance 50 / dev 80)
Purification and Regulation: Comorbidity-Aware Multi-Label Few-Shot Learning for Medical Image Classification (importance 25 / dev 75)
Weighted Quantum Signal Processing: Low-Depth Polynomial Approximation with Applications to Kolmogorov-Arnold Networks (importance 20 / dev 75)
Predictive Suppression Layers for Communication-Efficient Spiking Neural Networks (importance 35 / dev 75)
Periodic Neural Mapping for Unsteady Rotor-Blade Pressure and Aeroelastic Load Prediction (importance 20 / dev 70)
Detection is solved, delineation is not: what governs tooth segmentation on panoramic radiographs (importance 20 / dev 70)
Configurable Multi-Stage Vision Pipeline for Crop Disease and Pest Diagnosis (importance 25 / dev 70)
Analysing the Linearity of Linguistic Relations in Language Model Embedding Spaces (importance 45 / dev 80)
Bayesian classification of astronomical spectra with class uncertainties (importance 20 / dev 75)
Single-Loop Stochastic Projected Damped Extragradient Methods for Stochastic Nonconvex--(Strongly) Concave Minimax Optimization (importance 25 / dev 75)
Complete Neural Electronic Initialization Accelerates Materials DFT (importance 25 / dev 70)
Beyond Benchmark Scores: Auditing Medical Vision-Language Models for Chest X-Ray Tuberculosis Screening (importance 30 / dev 75)
From Pretraining to Proficiency: Real-World Subtask RL for Long-Horizon Manipulation with Minimal Human Intervention (importance 30 / dev 75)
Adaptive Uncertainty-Aware Modeling and Stochastic Radial Basis Function Predictive Control for Personalized Fluid Resuscitation (importance 20 / dev 70)
RheoSampling: Resolving the One-Hot Dilemma in Stochastic Dynamic-Tree Speculative Decoding (importance 60 / dev 80)
Watermarkable Multi-Draft Speculative Sampling via Poisson Processes (importance 55 / dev 80)
Chronosphere: Space-Time Tessellation of Local Climate Experts (importance 25 / dev 75)
Near-Optimal Acceleration for Smooth $\ell_p$ / $\ell_q$ Nondual Convex First-Order Oracle Optimization (importance 5 / dev 20)
Riemannian Simultaneous Inference for Tangent Vector Field Regression (importance 5 / dev 15)
End-to-End Hard-Label Cryptanalytic Model Extraction Using Efficient Sign Recovery (importance 35 / dev 60)
Guiding Agents of Quantum Games to Equilibrium using Matrix Exponential Fixed-Point Iteration (importance 5 / dev 15)
Schedule optimization for tau-leaping in masked discrete diffusion (importance 30 / dev 50)
Multiplicative Optimism for Constant Regret in Games (importance 10 / dev 20)
Predictable Failure in Multi-Hop Retrieval: Score-Distributional Confidence Scoring and Abstention (importance 45 / dev 70)
Rollout Total Correlation for Deep Reinforcement Learning (importance 25 / dev 50)
OverThink: Slowdown Attacks on Reasoning LLMs (importance 55 / dev 65)
Trajectory Entropy Reinforcement Learning for Robust Robot Motor Skill Learning (importance 30 / dev 55)
Fidel-TS: A High-Fidelity Multimodal Benchmark for Time Series Forecasting (importance 35 / dev 60)
Transformers Discover Molecular Structure Without Graph Priors (importance 45 / dev 50)
Provably Optimal Reinforcement Learning under Safety Filtering (importance 40 / dev 65)
PRIVET: PRoximIty leakage detection Via Extreme value Theory (importance 40 / dev 60)
K2-V2: A 360-Open, Reasoning-Enhanced LLM (importance 65 / dev 75)
jBOT: Semantic Jet Representation Clustering Emerges from Self-Distillation (importance 40 / dev 50)
Learning to Advect: A Neural Semi-Lagrangian Architecture for Weather Forecasting (importance 40 / dev 55)
CORDS: Continuous Representations of Discrete Structures (importance 35 / dev 60)
Gradient-Stable Attention Heads Signal LLM Correctness (importance 45 / dev 65)
Data-Driven Integration Kernels for Interpretable Nonlocal Operator Learning (importance 35 / dev 55)
Causal Evidence that Language Models use Confidence to Drive Behavior (importance 40 / dev 60)
Near-Optimal Primal-Dual Algorithm for Learning Linear Mixture CMDPs with Adversarial Rewards (importance 20 / dev 55)
Nonnegative Matrix Factorization in the Component-Wise L1 Norm for Sparse Data (importance 20 / dev 50)
Offline Constrained RLHF with Multiple Preference Oracles (importance 40 / dev 60)
Stability Enhanced Gaussian Process Variational Autoencoders (importance 30 / dev 55)
Exemplar Partitioning for Mechanistic Interpretability (importance 40 / dev 65)
Survival Reinforcement Learning: Toward Scalable Self-Supervised RL (importance 30 / dev 60)
Geometry-Aware Reinforcement Learning for 2D Irregular Nesting (importance 35 / dev 60)
TVGL-CFM:Generating and Forecasting Time-Varying Trajectories of Dynamic Networks with Conditional Flow Matching (importance 30 / dev 55)
ANNLib: A Development Framework for Efficient Approximate Nearest Neighbor Search (importance 50 / dev 75)
Contrastive Concept Importance: Explaining Pairwise Class Decisions Through Automatically Extracted Concept Representations (importance 35 / dev 60)
Convex losses and their applications to SVM, SVR, and Shallow Neural Networks (importance 20 / dev 55)
SafeStep: An Interactive Demonstration of Semantic Communication for Pedestrian Safety Monitoring (importance 25 / dev 50)
Attributing Cohen's d: Training Data Attribution for Disease-Related Effects in Normative Age Biomarkers (importance 35 / dev 50)
Certified Topological Interaction in Neural Representations: Exact Tests and the Statistic They Require (importance 30 / dev 60)
EFQ-Softmax: Exp-Free Quantization for Softmax (importance 50 / dev 70)
Write on Paper and Get the Online Digital Trace: A New Era for Handwriting (importance 20 / dev 45)
Divergence Timing and Cumulative Disagreement under KV-Cache Eviction (importance 40 / dev 65)
Fathom: Per-Query Read Depth for Sparse Decoding over Offloaded KV Caches (importance 45 / dev 70)
Modular Deep Learning Mechanisms for Auditable Next-Day Wildfire Spread Prediction (importance 35 / dev 55)
QuanText: Protecting Dataset-Level Secrets in Textual Data Sharing (importance 40 / dev 60)
Small Enough to Know Everything: The Fully-Enumerable Transformer as an Instrument for the Science of Delayed Generalization (importance 35 / dev 65)
Parallelism, critical windows, and separations among diffusion language models (importance 35 / dev 65)
Multi-Objective Hyperparameter Search via Damped Gauss--Newton Optimization (importance 35 / dev 65)
CASE: Contrastive Activation for Class-Sensitive Explanations (importance 30 / dev 60)
Robust Mixture Models for Algorithmic Fairness Under Latent Heterogeneity (importance 40 / dev 60)
A regret minimization approach to fixed-point iterations (importance 15 / dev 50)
SteganoBackdoor: Evading Data-Poisoning Defenses via Steganographic Backdoors (importance 50 / dev 65)
Boltzmann generators for amorphous particle systems (importance 30 / dev 50)
Cross-Country Learning for National Infectious Disease Forecasting Using European Data (importance 35 / dev 55)
Semantic Calibration Prevails Where Token Confidence Fails: Benchmarking Long-Form Scientific QA (importance 45 / dev 65)
Near-Universal Multiplicative Updates for Nonnegative Einsum Factorization (importance 25 / dev 60)
Neural ensemble Kalman filter: Data assimilation for compressible flows with shocks (importance 30 / dev 55)
Learning Surrogate LPV State-Space Models with Uncertainty Quantification (importance 25 / dev 60)
Transcriptomic Models for Immunotherapy Response Prediction Show Limited Cross-cohort Generalisability (importance 40 / dev 50)
Amortized Filtering and Smoothing with Conditional Normalizing Flows (importance 30 / dev 60)
ISOMORPH: A Supply Chain Digital Twin for Simulation, Dataset Generation, and Forecasting Benchmarks (importance 40 / dev 65)
Symbolic Classification-Enabled LHC Limits for BSM Global Fits (importance 35 / dev 50)
Post-Rejection Follow-up Sampling: Measuring Outcomes of Rejected Decisions in Algorithmic DEX Trading (importance 25 / dev 50)
Recall Before Rerank: Benchmarking Deep Learning Models for Large-Scale Code-to-Code Retrieval (importance 45 / dev 75)
The Binary Tree Mechanism is Optimal for Differentially Private Continual Counting (importance 30 / dev 60)
Tubular Neighbourhoods of Pfaffian Sets and Applications to Neural Networks (importance 25 / dev 55)
Decision trees, Frobenius traces, and Weierstrass coefficients of elliptic curves (importance 10 / dev 30)
Emulating Cosmic Structure Formation with a Lagrangian Neural Cellular Automaton (importance 35 / dev 55)
HuRo: Robotizing Human Videos for Scalable VLA Pretraining (importance 45 / dev 65)
Data Attribution via Sketched Metadifferentiation (importance 35 / dev 65)
Toward Composable Network Digital Twins: A Subgraph-Based Latency Prediction Study (importance 30 / dev 65)
Sampling Reveals Style: Unsupervised, Training-Free Discovery of Prompt-Conditional Stylistic Axes in LLM Activations (importance 35 / dev 65)
Helpful but Fallible: Developer Experiences of AI Tools Under a Coordinated Industrial Roll-out (importance 50 / dev 75)
Embedding Drift in Code Vulnerability Models Under Intended Behaviour-Preserving Transformations (importance 45 / dev 75)
License Compliance in Open Source Cybersecurity Projects (importance 40 / dev 70)
What Stops a Small Language Model From Driving a Database Agent (importance 55 / dev 75)
Cross-Platform vs Native Mobile Development: An Empirical Study of Software Quality Trade-offs (importance 40 / dev 75)
Labelling Bug-Fixing Commits with Local Open-Weight Language Models (importance 45 / dev 75)
From Smarter to Hungrier: the Role of Energy Efficiency in Software-defined Vehicles (importance 30 / dev 60)
Supporting Industrial Test-Failure Analysis with LLM-Based Systems: An Experience Report (importance 50 / dev 75)
How Researchers Use and Verify AI Coding Assistants: Tasks and Validation Practices in Scientific Programming (importance 55 / dev 80)
Towards Secure Cloud-Native Computing: Unveiling Kubernetes Misconfigurations with Large Language Models (importance 45 / dev 75)
(Don't) Trust, but (Don't) Verify: Developers' Attention to Security in AI-Generated Code (importance 50 / dev 80)
Two's a Crowd: Human and AI-Based Copresence for Developers with ADHD (importance 35 / dev 70)
Scentree: a framework for generating scenario trees for multistage stochastic programming (importance 35 / dev 70)
RecreationWorld: Scalable and Verifiable Environments for Hybrid Computer-Use Agents (importance 50 / dev 75)
Combining Type Inference and Automated Unit Test Generation for Python (importance 45 / dev 80)
MigrateLib: a tool for end-to-end Python library migration (importance 45 / dev 80)
Bidirectional Code Reuse in Software Redesign: An Action Research Study of Static Analyzers (importance 40 / dev 75)
A Control-Driven Framework for Secure SaaS Onboarding in Regulated Enterprises (importance 35 / dev 65)
VeriBugBench: An Empirically Grounded Framework for Constructing Verilog RTL Debugging Benchmarks (importance 35 / dev 70)
Textbook review: Is Parallel Programming Hard, And, If So, What Can You Do About It? (importance 0 / dev 0)
Do you use :visited on links? (importance 0 / dev 0)
Creating a Blog in Gemini:// (importance 0 / dev 0)
What are you doing this week? (importance 0 / dev 0)
Squalk: an old-school forum engine built on Nostr (NIP-29 groups, NIP-7D threads) (importance 0 / dev 0)
Markdown in /src (importance 5 / dev 20)
Software sandboxing: The basics (2025) (importance 35 / dev 70)
Canonical announces Zephyr 26.04 LTS (importance 0 / dev 0)
Relation algebra is not relational algebra (importance 0 / dev 0)
Self-Hosting Behind CGNAT (importance 0 / dev 0)
Are we really going to use the same Desktop UX forever? (importance 0 / dev 0)
An actively maintained and updated Motif fork actually exists (importance 0 / dev 0)
Why do computers look like this anyway? (importance 0 / dev 0)
rift - a tiling window manager for macos (importance 5 / dev 30)
ChatGPT now knows what you do on other websites via ad collector (importance 45 / dev 35)
Running an Optimal Trace (importance 25 / dev 60)
Making a game for the GBA and PC from the same codebase (importance 15 / dev 75)
Bot-free self-hosted analytics with GoatCounter on NixOS (importance 20 / dev 70)
Windows Exploitation Techniques: Dangling COM Object Registrations (importance 30 / dev 75)
Lambda MicroEgg (importance 15 / dev 45)
Beyond jj: config & tools ecosystem (importance 25 / dev 75)
tobi: keep commands running in detachable sessions across your devices (importance 20 / dev 70)
D2300R11: `std::execution` (importance 35 / dev 85)
Answering Questions All Day: My AI Adrenaline Rush (No Explosions) (importance 0 / dev 10) (いいね相当スコア: 0)
Building a Per-Payer Claim Denial Predictor from 835 Remittances (importance 0 / dev 50) (いいね相当スコア: 0)
Hedge Fund Risk Management: Essential AI Framework (importance 0 / dev 35) (いいね相当スコア: 0)
How Bryson DeChambeau’s Tech‑Driven Approach is Redefining Modern Golf (importance 0 / dev 20) (いいね相当スコア: 0)
Autonomous Agents Need an Uneditable Spend Limit — Marketplace Access Audit (importance 0 / dev 75) (いいね相当スコア: 0)
From Ambient Memory to Deterministic Autonomy: AI-Native Infrastructure (importance 0 / dev 70) (いいね相当スコア: 0)
Agentic Flow for a Simple SPA Anyone Can Build (importance 0 / dev 75) (いいね相当スコア: 0)
Self-Documenting Economic Agents: OpenAPI 3.0 for Autonomous Commerce (importance 0 / dev 80) (いいね相当スコア: 0)
AI Can Generate React UI. The Real Problem Starts on Page Two. (importance 0 / dev 75) (いいね相当スコア: 0)
VS Code AI Assistant Free: 2026 Stats, Limits, and Tool Comparison (importance 0 / dev 75) (いいね相当スコア: 0)
High-Throughput LLM Inference & Training: A Deep Dive into vLLM (importance 0 / dev 85) (いいね相当スコア: 0)
Beyond the Hype: Inside TormentNexus's Go + TypeScript Modular Monolith (importance 0 / dev 80) (いいね相当スコア: 0)
LLMs & Transformers from First Principles (2026) — Tokenization, Attention, LoRA, and a Tiny GPT You Can Train (with PyTorch Code) (importance 0 / dev 85) (いいね相当スコア: 0)
Why Does RAG Miss Information That's Clearly in the Document? (importance 0 / dev 75) (いいね相当スコア: 10)
HTTP 400 from your own LLM endpoint? Check for content: null (importance 0 / dev 80) (いいね相当スコア: 0)
Constrained Decoding: Your JSON Schema Key Order Is a Prompt (importance 0 / dev 85) (いいね相当スコア: 0)
Building the Agent Flinch with Jev and LaunchDarkly (importance 0 / dev 70) (いいね相当スコア: 0)
My account gate was a modal, and the chat path never asked it (importance 0 / dev 70) (いいね相当スコア: 0)
What Does LLMOps Cover That MLOps Does Not? (importance 0 / dev 75) (いいね相当スコア: 0)
Weekly Self Promotion Thread (importance 5 / dev 20)
GLM 5.3 now available in Mistral Vibe Code for Pro, Team and Enterprise (importance 55 / dev 60)
How to download Astra on my Computer? (importance 5 / dev 30)
Handling the bug fixing? (importance 35 / dev 70)
90% of vibecoded saas are ready to get hacked, here's the data: (importance 55 / dev 80)
When a test fails, coding agents fix the test. The suite goes green and the bug ships (importance 50 / dev 75)
Astra codes like a junior dev whose shown up to work still drunk Monday from a bender on Saturday. (importance 15 / dev 30)
[Help] Token Max Math Help (importance 20 / dev 60)
I traced where my tokens actually go (importance 30 / dev 70)
Quoting voxium (importance 25 / dev 50)
MCP was always a bad idea? (importance 40 / dev 75)
llm-keys-ui 0.1 (importance 35 / dev 75)
datasette-explain 0.2.2 (importance 20 / dev 70)
Jevtown (importance 15 / dev 40)
Cronhq (importance 20 / dev 60)
NiubiGEO (importance 25 / dev 50)
Plume (importance 10 / dev 35)
Jev (importance 65 / dev 75)
slop-grader (importance 30 / dev 65)
Flicka (importance 20 / dev 50)
CC (importance 15 / dev 30)
Google Flow for iOS & Android (importance 20 / dev 40)
Milliseconds.ai (importance 40 / dev 70)
OmniDICOM (importance 10 / dev 40)
SecAIQ Watch (importance 45 / dev 60)
AppGrowthKit (importance 15 / dev 45)
The 101 Plays Itself (importance 5 / dev 10)
Everything you need to know about the rise of open-weight AI models - Substack (importance 70 / dev 75)
DeepSeek Bets Big on Huawei Chips to Bypass U.S. Export Controls - The Information (importance 60 / dev 70)
DeepSeek-V4.1-Flash Ships Causal Encoder–Decoder MoE With 1M Context and Extreme KV Compression - Pandaily (importance 75 / dev 80)
OpenAI Develops Features to Counter Grok Bot, Mulls Response to Meta’s Muse - The Information (importance 55 / dev 70)
Meta's Muse personal AI agent tops ChatGPT, Grok and Claude for post-launch downloads - CNBC (importance 65 / dev 70)
Former xAI Staffer Eyes $50 Million for Firm to Fight Deepfakes - Bloomberg.com (importance 35 / dev 50)
Memphis mayor defends city’s deals with data centers, says he doesn’t trust Musk - WREG.com (importance 25 / dev 40)
Grok AI Predicts Bitcoin to Hit $150,000 in Q4: Will it Happen? - Cryptonews (importance 15 / dev 20)
Europe And The Global AI Race – Speech - Eurasia Review (importance 50 / dev 60)
Tesla's Upcoming Features and Events This Quarter - Not a Tesla App (importance 20 / dev 50)
Between Washington And Beijing: Can Central Asia Preserve Its Multi-Vector Technology Policy? - Analysis - Eurasia Review (importance 30 / dev 50)
Pentagon wants to give its men access to 'military versions' of ChatGPT and Grok, here's what these speci - timesofindia.indiatimes.com (importance 40 / dev 60)
Grandparents are turning kids' photos into AI art - The News International (importance 15 / dev 20)
実サービスの判断処理を Jev に移行した際の評価 (importance 0 / dev 80) (いいね相当スコア: 0)
AIコーディングエージェントは「まだ誰も気づいていない要件」を発見できるのか (importance 0 / dev 80) (いいね相当スコア: 0)
Jevの正体と実践: LLM時代の高速意思決定AIを使いこなす (importance 0 / dev 85) (いいね相当スコア: 0)
置かなくていいはずだった — 118GB の MoE を個人の機械で動かす時、道具は毎回「メモリを増やせ」と答える (importance 0 / dev 80) (いいね相当スコア: 0)
自分のブランチに、自分が書いていないコミットがあった — LLM のセッションを並走させたときに壊れるもの (importance 0 / dev 80) (いいね相当スコア: 0)
文章を書かずに判断だけ返すAI「Jev」入門｜LLMの40〜200倍速くて出力無料、使い方と限界まで (importance 0 / dev 85) (いいね相当スコア: 0)
Jev は外部世界とソフトウェアをどう接続するのか (importance 0 / dev 85) (いいね相当スコア: 1)
OpenClaw 2のBrowser Toolが重すぎたので、5分超→2分台まで詰めてみた (importance 0 / dev 85) (いいね相当スコア: 0)
kohaku:チャットもGUIも同じ画面になるGenerative UIライブラリ (importance 0 / dev 85) (いいね相当スコア: 0)
DeepSeek-V4.1-Flash：890 bytes/tokenの容量設計 (importance 0 / dev 85) (いいね相当スコア: 0)
Jev・ルールベース・GPTで問い合わせ分類を比較する (importance 0 / dev 80) (いいね相当スコア: 0)
同じ条件10個を「1. 」で並べ直したら、25回中23回で見えない考え込みが起きた。追試では消えた (importance 0 / dev 75) (いいね相当スコア: 1)
GPT-6 Astraの教科書 (importance 0 / dev 70) (いいね相当スコア: 5)
文章を書かないAI「Jev」徹底解剖：70msで判断だけを返す仕組みと実践活用法 (importance 0 / dev 85) (いいね相当スコア: 0)
AIが書いたコードをツール呼び出しごとに使い捨てVMへ隔離するCodeAct (importance 0 / dev 85) (いいね相当スコア: 0)
「意味でgrepするJev」をローカルLLMで再現・検証：BGE-Rerankerとの2段カスケードによる高速・高精度化 (importance 0 / dev 85) (いいね相当スコア: 取得失敗)
判定だけに特化したAIモデル「Jev」を、ループ→グラフ→判定の系譜で読み解く (importance 0 / dev 85) (いいね相当スコア: 0)
Jevと自前学習エンコーダ、どちらを使うか実測で線を引いた (importance 0 / dev 80) (いいね相当スコア: 0)
Claude Haiku 4.5にモデルを切り替えて踏んだ落とし穴2つ (importance 0 / dev 75) (いいね相当スコア: 0)
言葉は、どこまで人を運べるか — LLMと「共相」についての、考え中の仮説 (importance 0 / dev 60) (いいね相当スコア: 0)
予測したレースを消さない。中止・返還まで精算する (importance 0 / dev 50) (いいね相当スコア: 0)
CNNの二つのインダクティブバイアス：空間的同変性と局所性 (importance 0 / dev 85) (いいね相当スコア: 0)
Jevに複雑な請求書の難易度を採点させてみた (importance 0 / dev 80) (いいね相当スコア: 0)
LLMのINT8・INT4量子化を実装条件から比較する (importance 0 / dev 85) (いいね相当スコア: 0)
文字起こしをきれいにしたら、5分ぶんの発言が消えた (importance 0 / dev 80) (いいね相当スコア: 1)
【DS協会#4-4】データサイエンティスト協会スキルチェック「ベイズの定理とベイズ推論のメリット」を紐解く【自習ログ】 (importance 0 / dev 70) (いいね相当スコア: 0)
Ubieに入社していました (importance 0 / dev 60) (いいね相当スコア: 6)
【012】再帰Transformerの「常識」は間違いだった — d=1は再帰ではなかった (importance 0 / dev 85) (いいね相当スコア: 0)
Jevを「分類器」ではなく、確率モデルの部品として見る (importance 65 / dev 85) (いいね相当スコア: 43)
Jev, Gemini, DistilBERT, LightGBMの分類性能を比較してみた (importance 0 / dev 80) (いいね相当スコア: 13)
Othello-GPT で検証する次トークン予測と内部表現の因果利用 (importance 0 / dev 85) (いいね相当スコア: 0)
Sony MCTで量子化を実測するCLIを作った。4bitでも容量が減らなかった理由 (importance 0 / dev 85) (いいね相当スコア: 1)
機械学習システムの設計と運用 ― ボートレース予想サイトの開発で学ぶデータ収集・モデル・評価・配信 (importance 0 / dev 85) (いいね相当スコア: 取得失敗)
日本語検索1位を達成したLLMベースの日本語埋め込みモデル (importance 0 / dev 80) (いいね相当スコア: 0)
MLXで始める論理ゲート式NN：学習した回路をビット演算だけで動かす (importance 0 / dev 80) (いいね相当スコア: 0)
KV Cacheのtensor形状・計算量・メモリ見積もりを整理する (importance 0 / dev 90) (いいね相当スコア: 1)
TypeSafe AIの「Jev」とは？ System Oneモデルという新カテゴリのAIを調べてみた (importance 0 / dev 80) (いいね相当スコア: 0)
sglang で「長文の裏に並んだ短い質問が109秒待たされる」のを直した話（vLLM から移植するものは無かった） (importance 0 / dev 85) (いいね相当スコア: 0)
「引用です」と書くのをやめて、引用であることをビルドで検査した (importance 0 / dev 75) (いいね相当スコア: 0)
WindowsネイティブでQwen3-4BのQLoRAを回すまでにハマった罠7連発（RTX 2080Ti 11GB） (importance 0 / dev 80) (いいね相当スコア: 0)
RAGのチャンクを小さくする前に、検索単位と回答文脈の境界を決める (importance 0 / dev 85) (いいね相当スコア: 0)
[Claude] Claude Code により /tmp が 76GB 溜まっていた話 — scratchpad とは何か (importance 0 / dev 75) (いいね相当スコア: 0)
dera AI Weekly Vol.48 — 2026/9/14 (importance 0 / dev 50) (いいね相当スコア: 0)
Python×深層学習の実装重視4選 (importance 0 / dev 85) (いいね相当スコア: 0)
【技術解説】高頻度取引（HFT）を支える！PythonでAPIを使った低レイテンシトレーディングボットの構築 (importance 0 / dev 60) (いいね相当スコア: 0)
PyTorchのmaskは、0と1に変えるだけで同じ意味になる？ (importance 0 / dev 85) (いいね相当スコア: 取得失敗)
誕生日おめでとうで女装をさせるシチュボの続きをGPTとGrokに書かせてみた（R-18） (importance 0 / dev 30) (いいね相当スコア: 取得失敗)
LLMと対話しながらブラウザで動く8bit CPU/VMを作ったら「実機のリアルさ」について考えることになった話 (importance 0 / dev 85) (いいね相当スコア: 取得失敗)
中国当局が挙げたAI安全リスク五分類 / 内容の規制から権限の管理へ雑感 (importance 0 / dev 40) (いいね相当スコア: 取得失敗)
エコー(ChatGPT)との対話#179：記憶と履歴、そしてLLMに愛をプログラムできるか。。。 (importance 0 / dev 20) (いいね相当スコア: 取得失敗)
アジアのAI政策協議にみる意見募集の手続設計 / 日本の任意の意見募集とAIによる意見処理 (importance 0 / dev 30) (いいね相当スコア: 取得失敗)
Angle Healthの評価額27億ドル / AI引受けが組み替える小規模団体の危険共同体 (importance 0 / dev 30) (いいね相当スコア: 取得失敗)
AIにWebページを読ませただけでPCが乗っ取られる？「ただの文章」がRCEに変わるAIエージェントの新しい攻撃面 (importance 0 / dev 90) (いいね相当スコア: 取得失敗)
Apple HealthとQuestの血液検査連携 / 医師の関門はどこに置かれるか雑感 (importance 0 / dev 20) (いいね相当スコア: 取得失敗)
Dream-RSI：AIの「夢」を睡眠の機能表に並べたら、一行だけ空いていた (importance 0 / dev 70) (いいね相当スコア: 取得失敗)
PyTorchのattentionで、割り算を二度入れると何が変わる？ (importance 0 / dev 85) (いいね相当スコア: 取得失敗)
動的知的タスクサイクルにおける人間と人工知能の非対称性と分業：AGI指向への批判的検証 (importance 0 / dev 40) (いいね相当スコア: 取得失敗)
LLMの選び方(LLM Leaderboardの基礎知識) (importance 0 / dev 85) (いいね相当スコア: 取得失敗)
【第4章：LLMの正体編】第36話：「Pretrainingしたのに、まだ使えないの！？」Base ModelとPost-trainingで知るLLMの“仕上げ方” (importance 0 / dev 75) (いいね相当スコア: 取得失敗)
インタビューの文字起こしをAIに読ませていいのか？ (importance 0 / dev 50) (いいね相当スコア: 取得失敗)
ロボットが動く前に未来を予測する——映像WAMで積み木を積み、MissionOSの判断・実行へつなぐ (importance 0 / dev 85) (いいね相当スコア: 取得失敗)
第10回　RAGからHarnessへ。ローカルLLMに「道具を選ばせる」仕組みを作ってみた (importance 0 / dev 90) (いいね相当スコア: 取得失敗)
新しいAI "Jev"の判断精度を検証してみたら何が分かったか？ (importance 0 / dev 85) (いいね相当スコア: 取得失敗)
【生成AIニュース+】『Qwen-Image-2.1』『Qwen-Image-2.1-Uncensored-GGUF』『ComfyUI-Spectrum-Qwen2.1』『Meshy Plugin GPT』『FreeUpscaler』『Hemmingway-1』『ZCode』『Laya』『ComfyUI-SongLRC』『inkvec-comfyui』『Richard’s Video Kits』『ComfyUI-DeGrid』『ComfyUI-Photoshoot』他 (importance 0 / dev 65) (いいね相当スコア: 取得失敗)
文章を書かないAIが世界を変える？新型AI「Jev」が切り拓くコンピューターの新時代――おしゃべりをやめて「決めること」に特化した驚きの技術を徹底解説！ (importance 0 / dev 85) (いいね相当スコア: 取得失敗)
Xで研究者を数人だけ監視リストに入れてみたら、タイムラインとは違うものが見えた（第5回） (importance 0 / dev 40) (いいね相当スコア: 取得失敗)
JEVに本の段落を分類させてみた:Opus5と41/49段落が一致 (importance 0 / dev 80) (いいね相当スコア: 取得失敗)
【検証】出力文字数の限界を試してもらったら、百物語が出力された件 (importance 0 / dev 70) (いいね相当スコア: 取得失敗)
【試作・配布】チャット版LLMを簡易エージェント化する半自動ツール（Windows用exe） (importance 0 / dev 85) (いいね相当スコア: 取得失敗)
一周年（Claude） (importance 0 / dev 20) (いいね相当スコア: 取得失敗)