AI News

"Not" is Not Lost in Translation: A Causal Analysis of Negation Representations in Multilingual LLMs (arxiv.org)

Tokamak-SI: Interpreting Tokamak Discharges with Labeled Catalogs (arxiv.org)

Beyond Pattern Matching in 3D MLLMs: Part-Level Functional Reasoning for Generalizable Recognition (arxiv.org)

Representation Learning via Composite Subspace Flow Matching (arxiv.org)

Anti-coherence Regularized Absolute Scaling for Out-of-Distribution Detection (arxiv.org)

Easy-3Dfy: Object-Level Geometric Composition for Single-Image 3D Scene Construction (arxiv.org)

Head Dimension Matters in Language Models for Long-Context Retrieval (arxiv.org)

SOPO: Socratic Guided Policy Optimization for Span-Level Hallucination Detection (arxiv.org)

Dual-Adaptive Z-Score PPO: Dynamic Clipping Using Action Confidence and Normalized Advantage Magnitude (arxiv.org)

Uni-Stream: A Unified Streaming Speech Language Model for Speech Synthesis and Recognition (arxiv.org)

Searching with Sparse Histories: Cross-Task Architecture Search from Incomplete Search Logs and Zero-Cost Proxies (arxiv.org)

MG-REPA: Multi-Granularity Representation Alignment for Generative Target Speaker Extraction (arxiv.org)

Sparsity-driven emergence of functional modularity in multitask neural networks (arxiv.org)

Heterogeneous Space Data Centers: A Key Paradigm for Future Non-Terrestrial and In-Orbit Services (arxiv.org)

Beyond Data Schedule: Fast LLM Pretraining on Data Mixtures by Loss-Reweighting (arxiv.org)

ImageRSI: Verification-Driven Recursive Self-Improvement for Agentic Image Generation (arxiv.org)

FSVL: Taxonomic Fast-Slow Vision-to-Language Collaborative Framework for Interpretable Confusable Fish Species Identification and Knowledge Generation in the Middle Yangtze River (arxiv.org)

Machine-Readable Operator Manuals over the ISO 11783 Network for Offline, Multilingual LLM Assistance (arxiv.org)

SWE-Clar: Can Coding Agents Clarify Incomplete Client Requirements and Deliver Greenfield Software? (arxiv.org)

Rethinking Multi-View Fusion for Urban Representation Learning (arxiv.org)

SEVI: Safe Evidential Real-World Reinforcement Learning (arxiv.org)

How Ground Truth Shapes Retrieval Scores: Evidence-Aware Evaluation of a Retrieval-Augmented Generation System for Breast Lesions (arxiv.org)

Learning from Next-Frame Prediction: Towards Unified and Scalable Visual World Modeling (arxiv.org)

GeoDiv KAN Solver for Time Fractional Navier Stokes Equations on Complex Domains to Realistic CFD Reconstruction (arxiv.org)

Minimum Description Length Unsupervisedly Separates Law-Grade from Statistical-Grade Knowledge in a Self-Improving Science Agent - and Reveals 86% of Its Own Discoveries Are Statistical (arxiv.org)

What the Query Remembers: Query Abstraction Governs Lexical and Dense Retrieval of Classical Arabic Poetry (arxiv.org)

MusicTwins: Training Audio-Language Models to Listen like Their Persona (arxiv.org)

PrefO2O: Zero-Shot Preference-Based Reinforcement Learning via Offline-to-Online Latent Preference Transfer (arxiv.org)

Benchmarks Are Surprisingly Compaction-Friendly (arxiv.org)

Generalized green and red concepts for artificial intelligence, big data, and information and communication technologies: Multidimensional framework and countermeasures (arxiv.org)

Autoregressive Multimodal Anomaly Detection with Large Language Models for Microservice Systems (arxiv.org)

Do VLAs Know They Will Fail? Interpretable Success and Failure Patterns in Physical AI Systems (arxiv.org)

End-to-End Differentiable Diffusion Conditioning for Physics-Informed Optimization (arxiv.org)

What Open-Source Projects Tell Their Coding Agents: A Census of AGENTS.md and CLAUDE.md (arxiv.org)

Reuse, Rewrite, or Reject: Evidence-Constrained Trace Reuse for Repeated Mobile GUI Control (arxiv.org)

MBABench-2.0: Evaluating AI Agents on Constructing Financial Spreadsheets Beyond Simple Verifiability (arxiv.org)

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

BARC-Bench: Benign Asks, Risky Consequences (arxiv.org)

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

The Helpful Agent Problem: When Emergent Cooperation Among LLM Agents Bypasses Intended Workflows (arxiv.org)

Reproducible Evaluation as the Substrate for Safe Prompt Compression in Production Agents (arxiv.org)

Knowing Whom to Trust is not Enough: The Trust--Agreement Gap in Repeated Multi-Party LLM Negotiation (arxiv.org)

Beyond On-Policy vs. Off-Policy: At What Level Should LLM Agent Distillation Follow the Student? (arxiv.org)

Permission to Retry: A Study of Authorization and Recovery in AI Agent Workflows (arxiv.org)

TACS-RF: Tool-driven Agentic Circuit Synthesis of RF and mmWave Modules (arxiv.org)

Don't Forget What You Just Learned: Self-Directed Repair Turns Agent Mistakes Into Lasting Lessons (arxiv.org)

$M^3$: A Multi-LLM System for Multi-Turn Jailbreaking as a Markov Decision Process (arxiv.org)

Patronizing and Condescending Language Detection in English and Mandarin (arxiv.org)

Phronix: The Capability-Centric Learning Paradigm for Agents (arxiv.org)

Softly Symbolifying Kolmogorov-Arnold Networks (arxiv.org)