AI News

Conditional Flow Models for Offline Maximum Entropy Reinforcement Learning (arxiv.org)

Green Code or Costly Code? Evaluating the Energy Costs of Iterative LLM-Based Code Optimizations (arxiv.org)

A Foundation Model for Policy In-Context Longitudinal Causal Effect Estimation (arxiv.org)

Semantic Representation for Agentic Self-Play in Automated Laboratories (arxiv.org)

ENCL: Enabling Efficient Neuromorphic Continual Learning with Local Learning Rule and Sub-Network Freezing Strategy (arxiv.org)

From Deliberately Constructed Systemic Representations of the World of Activity to Dialogue: A Five-Level Communication Architecture for Strong Language-Based ASI (arxiv.org)

PolyMemDB: A Polyglot Database System for AI Memory Management (arxiv.org)

An Agentic Approach with Evidence-Grounded Verification for Multimodal Emotion Reasoning in Conversation (arxiv.org)

LeafSkill: Efficient Leaf-Level Action Grounding for Recursive LLM Agents (arxiv.org)

Do "True" and "False" Add Up to One? Complement Consistency in LLM-Agent Opinion Simulations (arxiv.org)

4DMemory: Benchmarking Long-Horizon Spatiotemporal Memory with Harness-Augmented Agents (arxiv.org)

PersonaScienceBench: Evaluating Scientific Agents with Diverse Simulated Researchers (arxiv.org)

CEMIR vs. FISER: a comparative study of reasoning-based and enrichment-based approaches to pragmatic intent recognition (arxiv.org)

TensorCommitments: A Lightweight Verifiable Inference for Large Language Models (arxiv.org)

Astro2Method: Benchmarking LLMs on Machine Learning Method Selection in Astronomy (arxiv.org)

Benchmarking Vision-Language Models on Synapse and Neuron Proofreading in Connectomics (arxiv.org)

MPQ: A Message-Passing View of Post-Training Quantization (arxiv.org)

DELA: Learning Controllable Edits for Diffusion Language Models (arxiv.org)

On-Policy Self-Training: Data-Free Self-Improvement in Large Language Models (arxiv.org)

Principia: Test-Time Hypotheses to Principle Construction for Scientific Discovery (arxiv.org)

Beyond Copying: Lifelong Smart Contract Security Task Construction via Protocol Mutation (arxiv.org)

AutoMIL: Automated Multiple Instance Learning for Whole-Slide Image Classification (arxiv.org)

Cycle-of-Thought: Self-Evaluative Iterative Reasoning for Table Visual Question Answering in Document Images (arxiv.org)

Rethinking LoRA Initialization for Robust Asymmetric Learning Rates (arxiv.org)

PPO vs. SAC: A Comparative Study for Cooperative Motion Control of Dual-Arm Space Robots (arxiv.org)

Validating the Comparison, Not Just the Scorer: Why Overall Accuracy Can Mislead in Interactive LLM Evaluation (arxiv.org)

When Clinical Judgment Becomes a Predictive Shortcut: Two Questions Clinical AI Must Answer Before Deployment (arxiv.org)

PaperFormalizer: Author-in-the-Loop Autoformalization of Research Papers in Lean (arxiv.org)

Learning Nonlinear Waveform Fusion Across Scales for Integrated Sensing and Communication (arxiv.org)

SteerVTE: Seamless Video Text Editing with Style and Glyph Control (arxiv.org)

$\Psi$-0.6: Adaptive Streams of Thought in a Unified Vision-Language World Model (arxiv.org)

Collaborative In-Context Learning for Indoor Radio Map Estimation (arxiv.org)

Monetizing Intelligence: Advertising and Subscription Design on LLM Platforms (arxiv.org)

US Wages Share of GDI at 75 Year Low (fred.stlouisfed.org)

Scaling FinOps Across Complex Google Cloud Environments: Enterprise Billing Agent (medium.com)

Behind the scenes at the NeXT Computer office (exhibits.stanford.edu)

PaperPilot – Zero-Cloud, Pure-Rust PDF Engine with WASM and MCP (github.com)

Show HN: TitleArt – Make text like it's 1997 (titleart.dvy.io)

Microsoft Decision-1: 9B Qwen Model Launch Explained (shattered.io)

Brazil elections: Lula secures fourth term after dramatic comeback (www.bbc.com)

L'État mobilise 7 milliards de France 2030 pour accélérer les projets industriels en Auvergne ... (renseignementeconomique.fr)

Standfast (www.producthunt.com)

If AI displaces human workers, they will lose more than just pay (www.economist.com)

Google ADK Training 2.0: The Definitive Guide to Mastering Production-Ready Multi-Agent Systems (medium.com)

IBM CEO says open weight models are now irresistibly good and cheap on-prem (www.theregister.com)

25% of Incoming Pull Requests May Be AI Slop, Finds Study From Linux Foundation Research (news.slashdot.org)

iPhone Duo Teardown (www.ifixit.com)

Singapore-linked data center projects face growing rejection in Indonesia - Nikkei Asia (asia.nikkei.com)

Removing Next.js Got My Blog to 100 on Every Lighthouse Category (korindiehacker.bloffy.blog)

Lula wins election after final votes tallied (www.reuters.com)