AI News

Attention-Mass Condensation for Sparse Decoding (arxiv.org)

DiTS: Multimodal Diffusion Transformers Are Time Series Forecasters (arxiv.org)

Democratizing MoE inference on commodity GPUs with CoMoE (arxiv.org)

Scaling Down the Scaling Laws: Parameter Efficiency and Compute-Optimal Training in Resource-Constrained Large Language Models (arxiv.org)

PIT-GCL: Protein Interaction using Topological Graph Contrastive Learning (arxiv.org)

Walk fast but be careful: Understanding Parallel Sampling in Masked Diffusion (arxiv.org)

Train4Merge: A Controlled Single-Teacher Study of RL vs. SFT Teachers for OPD-Based Model Merging (arxiv.org)

Residual-Stream Burden Shapes Representation Learning in Diffusion Transformers (arxiv.org)

Multi-LLM Collaborative Alignment via Stackelberg Games (arxiv.org)

Questioning the Questions: Sustaining Self-Evolution in Reasoning Models (arxiv.org)

Humanoid Rickshaw Pulling: Whole-Body Locomotion under Coupled Wheeled Loads (arxiv.org)

The sublevel Flood bifiltration: towards scalable 2-parameter persistent homology (arxiv.org)

RAISED: Self-Distillation for Robustness to Prompt Injection in LLM Agents (arxiv.org)

Boosting Transferable Adversarial Attacks against Deep Reinforcement Learning (arxiv.org)

OpenTSLM TeeMoE: A Unified Time-Series Language Model for Forecasting, Contextual Prediction, and Reasoning (arxiv.org)

Clean: Second-order LLM Training at Linear Memory Cost via Nystr\"om Sketching (arxiv.org)

Boosting and the Expressive Power of Simple Weak Learners via the $\gamma$-VC Dimension (arxiv.org)

ProtocolMatch: Protocol-Dependent Model Selection for Scientific Dynamics Forecasting (arxiv.org)

From Prompts to Trees: Effective LLM-Guided Tree Generation for Few-Shot Tabular Classification (arxiv.org)

Evaluating Sequence Assembly Strategies for Differentially Private Synthetic Time-Series Forecasting (arxiv.org)

Edge Accuracy Is Not Enough: Why Dynamics-Learned Structure Fails to Transfer to Inverse Problems (arxiv.org)

OrthoGen: A Generative Orthogonal Learner for Time-Varying Treatments (arxiv.org)

AutoAdapt: Automatic Domain Discovery Enables Low-Cost Extensibility (arxiv.org)

How to train your model organism (arxiv.org)

Robust Decentralized Fairness Auditing (arxiv.org)

A Unified Information-Theoretic Approach to Constrained Multi-Fidelity Multi-Objective Bayesian Optimization (arxiv.org)

What Can a Gaussian Process Design Test (arxiv.org)

CAFE+FNO: Fourier Kernel Generation via Multiplicative Feature Composition (arxiv.org)

Decoupling Exploration from Optimization in RLVR (arxiv.org)

From Plausible Hierarchies to Useful Taxonomies: Evaluating Agentic Harnesses on Customer Feedback (arxiv.org)

Multi-Agent Coordination via Support-Preserving Distillation (arxiv.org)

Activation-Aware Weight Tensorization: A Calibration-Time Preconditioner for Tensor-Network LLM Compression (arxiv.org)

Efficient Provably Private Classification with a Tabular Foundation Model (arxiv.org)

WxFM-XL: Adapting Univariate Foundation Models to Multi-Station Weather Forecasting (arxiv.org)

Many Ways to Succeed: Diversity-Driven RL Fine-Tuning for VLA Generalization (arxiv.org)

Evolutionary Architecture Search for Chlorophyll-$a$ Prediction in Lakes using Sentinel-2 (arxiv.org)

Conditional Flow Matching for Generation of 3D Multi-variable Instantaneous Urban Microclimate Fields (arxiv.org)

Steerspeech: Activation Steering For Emotion Control In Generated Speech (arxiv.org)

YANchor-4B: Effective Long-Horizon Reasoning in O(N) Time with O(1) Memory (arxiv.org)

RobotWorld: Benchmarking Multimodal Agents for Robot Use Across Diverse Tasks and Embodiments (arxiv.org)

Rubix: Global Correspondence-Free Point Set Alignment through Assignment Geometry (arxiv.org)

SOTA: Stock Options Trading Agents Guided by Option-Implied Return Distributions (arxiv.org)

Pathwise Information Certificates for Decentralized Adaptive Sensing (arxiv.org)

Measurement-Efficient Differentiable Quantum Architecture Search for Combinatorial Optimization (arxiv.org)

Estimating Uncoded Crash Factors with Tabular Foundation and System One Models: Kumo Tabular and Jev (arxiv.org)

Energy-Efficient Gait Adaptation via Hierarchical Reinforcement Learning for Quadrupedal Locomotion Across Diverse Terrains (arxiv.org)

AI Safety Considerations for Agents With Limited Time to Act (arxiv.org)

Temporal Visuo-Tactile Learning for Dexterous Grasp Stability (arxiv.org)

CARES: A Controlled Synthetic Benchmark of Speaker Reactions to Sound (arxiv.org)

GAGR-Lab: Evaluating Joint Spatial-Geometric and Analytic Function Reasoning (arxiv.org)