Signal over volume.

ComBench: A New Benchmark for Olympiad-Level Combinatorial Reasoning and Cons...

ComBench: A New Benchmark for Olympiad-Level Combinatorial Reasoning and Construction

RACES: Scaling Reasoning via Recursive Composition of Verifiable Environments

Arbor: A General Framework for Autonomous Research via Hypothesis-Tree Refine...

DeNovoSWE: A Large-Scale Dataset for Scaling Long-Horizon Repository Generation

Platform Sorting Drives Ideological Fragmentation in the Social Media Ecosystem

Algorithmic Delivery of Political Ads Skews Toward Male Audiences for Populis...

Pareto-Guided Teacher Alignment for Fair Personalized Text Generation

Classroom Study Reveals Natural Language LLM Feedback Outperforms Test Case F...

PPC Learning Rate Selection Fails in Gaussian Models Due to Data-Free Degeneracy

RETROSPECT: A Modular Proposal-and-Reranking Framework for Single-Step Retros...

The Identity Trap: Diagnosing Subject-Identity Confounding in EEG Foundation ...

Human Developers Fail to Detect AI Agent Sabotage in 94% of Long-Horizon Codi...

GDPRuler: Verifiable GDPR Compliance for Key-Value Stores via Trusted Middleware

VISMATIC: A Secure Rootless Framework for Monitoring Student Coding Processes...

Partisan Persona Prompting Increases Political Polarization and Persuasion En...

UK Online Safety Act Triggers Surge in VPN Interest and Privacy Concerns

Physics Is All You Need? Why AI Agents Need Human Physicists to Avoid 'Fudge ...

Generative Protein Models Outperform Traditional Sampling in Mapping Conforma...

IndicBERT-HPA: Reliable Multilingual Orthopedic Decision Support for Low-Reso...

Quantum Element-wise Transforms: Exponential Space Reductions for Matrix Func...

Trapped-ion quantum computer demonstrates breakeven performance with high-rat...

Robustness of Entanglement Manipulation for MSR Almost i.i.d. Sources

Absence of the Mpemba Effect in Boundary-Driven Free Fermion Thermalisation

Introducing Repeated Policy Regret (RP-Regret) for Games with Adaptive Opponents