AI News

A Language-Informed Loss for Visual Generation (arxiv.org)

Scalable long-range machine learning interatomic potentials with electrical response for atomistic material modeling (arxiv.org)

Trajectory-Triggered Sparse Collaborative Decoding (arxiv.org)

An AI Approach to Verified Production Cryptographic Libraries (arxiv.org)

Every Model Cheats: Prompt-Level Mitigation of Cheating on Offensive Cyber Tasks (arxiv.org)

FaithfulFaces: Pose-Faithful Facial Identity Preservation for Text-to-Video Generation (arxiv.org)

VideoHarness-SRSI: Structured Recursive Self-Improvement for Video Understanding Harness (arxiv.org)

Interpreting SSL Representations for Spoof Detection: a WavLM Study (arxiv.org)

Graphlet-based Decentralized Finance Transaction Method Classification (arxiv.org)

HydraLLM: Adaptive Width Slicing for Elastic Text and Vision-Language Models (arxiv.org)

Sentence-Level Topic Modeling with Lexical and Contextual Evidence (arxiv.org)

Data-Driven Prediction of Optimal MPI Rank Placement and Node Count for Memory-Bound CFD on AMD EPYC Clusters (arxiv.org)

GroundingPI: A Grounding Foundation Model towards Physical Intelligence with Visual Primitives (arxiv.org)

Beyond Success Rate: Interventional Evaluation of Memory in Vision-Language-Action Systems (arxiv.org)

EvalResearchBench: Can AI Agents Build Their Own Feedback for Recursive Self-Improvement? (arxiv.org)

Trust Is Not a Score: Runtime Assurance Contracts for High-Risk AI Agents (arxiv.org)

When Should Agents Check External State? Budgeting Observations for Stored Intentions (arxiv.org)

Action-Space Shaping for LLM Agents: Measuring and Mitigating Tool-Schema Bias (arxiv.org)

SpeechCritic: Learning a Diagnostic Speech Judge from Limited Human Preferences (arxiv.org)

R$^2$ Flow: Recursive Self-Improvement via Recursive Skill Evolution (arxiv.org)

RSD-Poker: Structure-Adaptive and Shift-Robust Risk-Utility Certification for Residual Policies in Imperfect-Information Games (arxiv.org)

Statistical Vignettes on the Grokking Phenomenon (arxiv.org)

SP$^2$ec: Adaptive Self-Speculative Decoding for Vision-Language Models (arxiv.org)

LoongSage: A Production-Grade, High-Efficiency Agentic RL Framework for Frontier LLMs with Validated Recipes (arxiv.org)

Audit-First VAPO: Risk-Certified Selective Updates under Imperfect Verification (arxiv.org)

OSCC: Certified Observation-Safe Coupling Optimization for Gradient-Noise Control in Imperfect-Information Learning (arxiv.org)

Porimon: An LLM-Based Pok\'emon Battle Agent Enhanced by Long/Short-Term Knowledge Augmented Generation (arxiv.org)

Metro-WM: Long-Horizon Latent Planning with Realisable Sub-Goals (arxiv.org)

Error Correction in Trained TTT Layers (arxiv.org)

Guarded Block Transfers for Polytope Distance: Linear Convergence with Explicit Constants and Gap-Certified Screening for SVM Training (arxiv.org)

Look, Reason, Draw a Region: Recursive Refinement for Spatial Context Prediction (arxiv.org)

How the elastic axis folds the transonic flutter boundary: shock energy transfer and conformally calibrated active learning (arxiv.org)

Learning angles for quantum approximate optimization (arxiv.org)

The Optimal Ranking Tradeoffs Between Two Ranking Scores in Two-Class Classification (arxiv.org)

Distributed Performance Boosting through Neural Control for Plug-and-Play Networks (arxiv.org)

NestedVLA: Learning to Consolidate and Generate Skills for Vision-Language-Action Models (arxiv.org)

Can the Parts Fool the Test? Counterfactual Pairing Cycles for Relational OOD (arxiv.org)

Think More or Read More? A Token-Budget Theory of Reasoning with Shared Context (arxiv.org)

Proportionality in Ranking Compression (arxiv.org)

DIFFLET: A Fast Inference Engine for Diffusion Transformer Models on AWS Trainium (arxiv.org)

Controlling the Diversity-Quality Trade-Off of LLMs by Merging Models Before and After Alignment (arxiv.org)

How Murky Before a VLM Knows It Cannot See? Accuracy, Confidence and Abstention on a Physical Visibility Axis (arxiv.org)

What Shapes Brain Alignment of Large Language Models during Dialogue? Insights from Language Production and Comprehension (arxiv.org)

COMET: Decoupled Distillation, Routing, and Capacity Control for Task-Agnostic Continual Vision--Language Learning (arxiv.org)

Beyond Uncertainty Reduction: Calibrated Planning for Interactive Diagnosis (arxiv.org)

CDAE: Enhancing Perturbation Robustness in Pretrained Language Models with Contrastive Denoising (arxiv.org)

Where's the Indirect Evidence? Investigating Grammatical Generalization with Influence Functions (arxiv.org)

FederatedBytes: Tokenizer-Free, Byte-Level Small Language Models for Efficient Multimodal Federated Learning (arxiv.org)

Competence-Based Evaluation of Logical Reasoning in Large Language Models (arxiv.org)

HAPACT: A Benchmark For Human-Centric Physical Impact Localization in Movies (arxiv.org)