AI News

SPARTA: A Sparse Ternary LLM Accelerator with Adaptive Attention and Unified Datapath for Long-Context Edge Inference (arxiv.org)

Rectified Language Flow (arxiv.org)

Silent and Costly: Characterizing and Detecting Prompt-Caching Mis-Integration in LLM Agents (arxiv.org)

Precedence-Aware Resource Allocation: Extending the AUTOC Framework to Multi-Level Treatments (arxiv.org)

ToDoPro: Faster and Better Reference-Based Generation: Early Token Reduction via Downsampled Self-Probing (arxiv.org)

The Impact of Personalized AI Response Suggestions on Content Creators' Direct Messaging (arxiv.org)

Curvature-Aware Clipping: Mitigating Bias in DPSGD via Approximate Second-Order Information (arxiv.org)

Putting Transformer Theories to the Test (arxiv.org)

PRISM: Promoting Sampling Diversity of Large Language Models with Implicit Stochastic Modulation (arxiv.org)

GS-MoE: Training-Free MoE Compression via Groupwise Subspace Sharing and Pruning (arxiv.org)

REDAR: Reinforcement-Driven Adaptive Regularisation for Class Incremental Learning (arxiv.org)

DynEdit: Dynamic Entropy-Guided Sequential Editing for Large Language Models (arxiv.org)

The Physical Cost of Context: Quality-Constrained Resource Accounting for Coding Agents (arxiv.org)

Babysitting AI Agents: A User-Centered Taxonomy and Study of Agentic Delegation Failures (arxiv.org)

SceneAgent: An Agentic Pipeline from 3D and 4D Captures to Simulation Environments (arxiv.org)

VLA-HWBench: Where the Time Goes When VLA Models Run on Edge Hardware (arxiv.org)

Beyond Heuristics and Commercial Navigation: An AI-Augmented Decision-Support for Ambulance Travel Time Estimation (arxiv.org)

Sharp Bounds on the Winner's Bias under Independent and Nonnegatively Correlated Noise (arxiv.org)

Towards Trustworthy AI-Enabled Emergency Medical Dispatch: Uncertainty-Aware Ambulance Travel Time Estimation (arxiv.org)

Agreement rather than normalisation keeps depth reusable in looped language models (arxiv.org)

Beyond Local Accuracy: Benchmarking the Rollout Reliability of PDE Foundation Models (arxiv.org)

Robust and Fine-Grained Detection of AI Generated Texts (arxiv.org)

Gangqinpu: A Multi-Genre, Multi-Task Piano Dataset with Aligned Audio, MIDI, and Musical Scores (arxiv.org)

From Speech Tokens to Stolen Voices: Attacker-Chosen-Text Speaker Impersonation via Representation Inversion (arxiv.org)

Knowing When the Evidence Is Enough: Requirement-Directed Search with Horn-Closure Certificates (arxiv.org)

Benchmarking Obstacle-Conditioned Safety in Household Robot Navigation (arxiv.org)

Agent Memory as Recommendation: Recommender-System Design for Long-Term Memory of LLM Agents (arxiv.org)

Video World Models Need Predictable Latents (arxiv.org)

Thinking-in-Bets at Test Time Reduces Sycophantic Revisions in LLM Agents (arxiv.org)

RobotAgentBench: Towards Open-world Robot Autonomy (arxiv.org)

Every Weight Is a Path of Shared Discrete Decisions (arxiv.org)

Decode Late, Multiply Once: Rethinking the Dot Product for Quantized Weights (arxiv.org)

Topological Equivalence of Spectral and Density Cluster Hierarchies (arxiv.org)

Scaling Transformer Energy-based Models For High Quality Image Synthesis (arxiv.org)

Forecasting Associations, Not Interventions: A Preregistered Causal Audit of Covariate-Aware Time Series Foundation Models (arxiv.org)

Rewriting Scaling Laws for Synthetic Data with Compute and Data Constraints (arxiv.org)

Intrinsic Goals for Autonomous Agents: Hierarchical Exploration Captures Future Volitional Action in Drosophila (arxiv.org)

AgentProofBench: Benchmarking Reusable Verifiers for Web Agents (arxiv.org)

CodeMeepo: A Multi-Agent Workflow for Repository-Level Code Migration and Optimization (arxiv.org)

RIDE-Splat: Reflection Removal for In-the-Wild Driving Environments (arxiv.org)

Entity Matching using Decision Models (arxiv.org)

Aligning Large Language Models with Human Feedback: Mathematical Foundations and Algorithm Design (arxiv.org)

Cautious Conflict-Free Inverse Gradient Methods (arxiv.org)

Speculative Diffusion is a Tale of Two Tradeoffs (arxiv.org)

Nested Sparsity for Scalable Pretraining and Continual Learning (arxiv.org)

CuraLoc: Repository-Level Fault Localization via Planned Code Inspection (arxiv.org)

Unlocking the Specialization Capability of PPG Foundation Models for Wearable Signals through Structured Pruning and Adaptive Fine-Tuning (arxiv.org)

InfiniteWorld: Separating Visual Evidence from Familiar Physics (arxiv.org)

The Hitchhiker's Guide to System 1 Models: Fast Decision-Making Beyond LLMS (arxiv.org)

An actively trained neural operator for near-explicit atmospheric chemistry (arxiv.org)