Trending

Recent advances in post-training Large Language Models (LLMs) increasingly rely on Reinforcement Learning with Verifiable Rewards (RLVR) or On-Policy Self-Distillation (OPSD). While OPSD provides dense, logit-level supervision, it…

Perovskite oxides have emerged as an important class of material with promising energy applications owing to their compositional and structural flexibility, which enables stabilization of both low- and high-symmetry phases and gives rise to…

Many properties of classical graphs are defined in terms of subsets of the vertex set. Examples include connected components, which are subsets $X \subseteq V(G)$ such that $X \times X^c$ and $E(G)$ are disjoint, or independent sets, for…

AI accountability at scale is an institutional problem: who can observe, verify, and change deployed systems. We develop a sequential political-economy model in which an AI vendor chooses auditability and substantive mitigation, a deployer…

Large language model training in open-ended domains lacks verifiable rewards, making task preferences difficult to formalize as effective supervision. Contexts can convey such preferences, yet provide little additional supervision once…

By examining the relationship between the support function of convex geometry and the Euler Characteristic Transform (ECT) of topological data analysis, we develop new tools and suggest variations on some common ECT pipelines. Specifically,…

Learned video compression relies on accurate temporal modeling to remove redundancy between adjacent frames. However, most existing codecs infer motion solely from discretely sampled RGB frames, making their estimates vulnerable to fast…

User foundation models have demonstrated strong results in e-commerce and social recommendation, but most industrial deployments assume environments where user identity is stable and persistent. Open-web real-time bidding (RTB) operates on…

Reconfigurable intelligent surfaces (RISs) are a promising technology for improving the spectral and energy efficiency of future wireless networks, which make use of metasurfaces. However, optimizing RIS configurations typically leads to…

Chimera states, characterized by the coexistence of synchronized and desynchronized dynamics in identical oscillators, are typically studied in systems with pairwise interactions. Whether higher-order interactions alone can generate such…

In 2012, Kaufman and Lubotzky constructed the first family of symmetric LDPC good codes. Their construction used Cayley codes, as originally defined by Kaufman and Wigderson (2016). In this paper we present two generalisations to the Cayley…

Bayesian posterior sampling is a ubiquitous paradigm for problems where a point estimate of parameters is not sufficient, such as risk analysis and uncertainty quantification. However, likelihoods may be misspecified, intractable,…

Although neural-based machine learning models have received a lot of attention recently, tree-based models such as gradient boosting are competitive for tabular data and therefore remain widely used in various applications of AI. As when…

One might expect the most massive stars to also be the largest by size, yet they are not, and this has puzzled astronomers for decades. The Eddington limit sets an upper bound to stellar mass through the balance between radiation pressure…

The evolutionary pathways and ultimate fates of very massive stars are governed primarily by mass loss through radiatively-driven winds. We present a new theoretical mass-loss prescription for (very) massive stars, capturing the complex…

We give a homological proof of the famous Gorenstein Symmetry Conjecture, which asserts that one-sided finiteness of the self-injective dimension of an Artin algebra is sufficient for Gorensteinness.

Novel findings on nanostructures in semi-metal and semi-graphene materials are discussed regarding rectangular quantum dots with zigzag edges, MIT bag models, and Dirichlet boundary conditions. The article theoretically investigates some…

We describe infinitesimal deformations of post-Lie algebras and post-Hopf algebras and prove that the adjunction given by the universal enveloping algebra and primitive elements functors is compatible with the infinitesimal structure. When…

Representation engineering reads and steers capability directions in large language models, yet methods are typically evaluated on paper-specific synthetic data. The resulting measurements are difficult to compare or reproduce and may…

Pathology foundation models (FMs) are models trained on vast amounts of typically unlabeled data and have been shown to yield regularized latent spaces that can be used effectively in downstream classification tasks. This is also true for…