Leaderboards, eval harnesses and ablations — the contested business of deciding which model is actually better.
Where generalization capacity--the ability to extract context-invariant relational structures--first emerges remains a central question in AI and neuroscience. The foundation for this capacity lies up...
As LLMs increasingly assist in moral reasoning, omission bias, the tendency to prefer inaction even when equivalent framings reverse substantive outcomes, poses a significant risk of skewed decision-m...
Multi-agent systems (MAS) powered by large language models have shown strong performance across code generation, mathematical reasoning, and question answering. However, existing methods for automatin...
In this paper, we show that standard evaluations of high-resolution Model Inversion Attacks (MIAs) significantly underestimate training-data privacy leakage. State-of-the-art privacy defenses, standar...
Out-of-distribution (OOD) generalization remains challenging for graph neural networks (GNNs), as graph distributions can vary substantially across time and domains. Supervised and self-supervised gra...
We evaluate whether AI agents can match the reported findings of published biomedical studies using public data. Existing evaluations do not consistently separate analysis from prior knowledge or retr...
In clinical practice, agreement among independent experts is treated as evidence of reliability, and multi-round consensus has become a core mechanism of agentic medical question-answering systems. Wh...
Robust person re-identification often combines complementary cues such as face, gait, and body shape. While adaptive fusion typically targets query quality, model strength also varies across identitie...
T-Search is an open-weight agentic retriever designed for hard multi-step search tasks. It performs bounded multi-round searches over a fixed corpus, returning ranked evidence chunks with brief justifications, while allowing the answer generation component to be swapped without retraining. Trained on synthetic search tasks and evaluated on seven English and Russian benchmarks, it achieves 56.0 Recall@10 with one rollout and 61.3 with three fused rollouts, surpassing larger open models.
The paper explores how prior-data fitted networks (PFNs) can be adapted for anomaly detection in tabular data. Using frozen TabPFN features, the authors show that scoring samples by nearest-neighbor distance in feature space yields strong results, and they identify optimal layers and extraction procedures. They further improve performance by fine‑tuning the model on a reference set, achieving higher AUROC on the ADBench benchmark with both a fine‑tuning‑free (ZEN) and a fine‑tuned (FOCUS) approach.
The paper highlights that AI voice assistants using ASR and LLMs struggle with regional British accents because most ASR models are trained on American English. It introduces CavaBench, a benchmark of spoken financial queries, to evaluate ASR models and their impact on downstream tool‑calling accuracy across British accents. The study finds that while WER predicts tool‑calling accuracy, it may not fully capture task‑level performance, revealing accent‑related failures that vary by model and acoustic conditions.
The paper investigates how frontier AI systems perform on tasks related to cyber security, specifically exploit generation, vulnerability repair, and subsequent attacks, using five nonpublic software environments. It evaluates both open‑weight and proprietary models, employing deterministic graders rather than LLM judges to score performance. Results show significant variation across systems and vulnerability types, with repair scores higher than attack scores in two environments and lower in three, and highlight that passing an initial security test does not guarantee long‑term defense, as 92 of 524 defender test intervals still experienced successful exploits after the first one was stopped.
MTOR introduces a generalizable AI‑generated video detector that combines global visual features with caption‑derived textual semantics and a novel Temporal Over‑Regularity (TOR) component. The TOR module captures three levels of temporal consistency—coarse inter‑frame continuity, fine‑grained token correspondence, and frame‑to‑video stability—to exploit the stronger temporal persistence and lower variability found in synthetic videos. Extensive tests on five benchmarks with 46 generator variants show MTOR outperforms 16 baselines and remains robust against twelve real‑world video perturbations.
The paper introduces VERA, an automated framework that audits large language model (LLM) reasoning in software vulnerability analysis. Instead of relying on free‑form explanations, VERA requires models to produce a Structured Reasoning Record (SRR) that captures pointers, memory operations, and state transitions in machine‑readable fields. A multi‑stage judge then checks each SRR against eight reasoning failure modes, revealing that reasoning flaws are as common in correct verdicts as in incorrect ones and that VERA detects 87% of errors missed by free‑form LLM‑as‑judge evaluations.
GraphDecide is a model‑independent benchmark designed to evaluate System One models—such as Jev—that make decisions directly from supplied options on graph‑related tasks. The benchmark combines structural task profiles, matched graph‑text input contrasts, and heuristic‑proposal controls to diagnose graph decision performance. In testing fourteen model‑interface configurations, GraphDecide shows that accurate adjacency recognition does not guarantee broader structural correctness, joint graph‑text input does not consistently improve prediction, and feasible construction does not ensure high solution quality.
The paper introduces Abusive Language Identification (ALI), a sequence‑labeling approach that simultaneously extracts abusive spans and identifies target mentions, addressing limitations of traditional sentence‑level abusive language classification (ALC). Using a pilot corpus from a production moderation pipeline, the authors compare ALI to ALC, finding that ALI performs competitively while offering localized outputs useful for moderators, with a modest advantage on implicit abuse. However, accurately determining exact abusive boundaries and target spans remains challenging, and the study highlights the need for further research on target‑span linking and structured benchmarks for ALI.
The paper introduces Guidance‑TTT, a method that separates strategic planning from execution in test‑time training for large language models. A small guidance model is trained at test time to propose high‑level changes, while a frozen, larger execution model implements these changes, reducing the cost of maintaining gradients and optimizer states. Guidance‑TTT achieves strong results across four domains—combinatorial optimization, heuristic programming, machine learning, and GPU kernel optimization—outperforming prior work and matching state‑of‑the‑art leaderboard scores.
The paper introduces a dataset‑free approach to generate planar graph embeddings that meet strict structural and geometric constraints. By integrating parametric graph grammars with safe reinforcement learning, the authors formulate the task as a constrained Markov decision process and use action projection to keep actions within safe sets. Their method consistently outperforms classical and deep generative baselines on a new benchmark suite for constrained, goal‑directed planar graph generation.
The paper introduces PIVOTMIPL, a method for video classification that tackles multi‑instance partial‑label learning by jointly assigning classes and temporal evidence. It uses occupancy‑regularized spherical matching to link contextualized video features with nonuniform temporal mass, and a dual‑marginal KL projection to build a structured teacher that refines class beliefs while maintaining temporal occupancy. Experiments on VCMIPL benchmarks derived from Breakfast, DoTA, and FineAction show that PIVOTMIPL outperforms existing MIPL algorithms in both effectiveness and efficiency.
MoCAR (Motion-code Coordinate-aware AutoRegression) is a decoder‑only framework that treats continuous trajectory forecasting as next‑code prediction in a coordinate‑aware latent space. It learns a continuous motion‑code space from endpoint‑normalized trajectory segments, using historical codes as a teacher‑forced prefix and generating future codes autoregressively while updating local scene context. The method achieves top‑tier performance on Argoverse benchmarks, transfers well from AV2 to AV1 in zero‑shot evaluation, and improves on turn‑heavy scenarios without requiring trajectory‑space re‑tokenization or proposal‑and‑refinement pipelines.