Leaderboards, eval harnesses and ablations — the contested business of deciding which model is actually better.
arXiv:2603.16017v2 Announce Type: replace-cross Abstract: Large language models (LLMs) increasingly participate in morally sensitive decision-making, yet how they organize ethical frameworks across r...
arXiv:2603.23184v2 Announce Type: replace-cross Abstract: Despite the success of reinforcement learning from human feedback (RLHF), existing reward modeling methods largely rely on explicit feedback,...
arXiv:2605.18882v2 Announce Type: replace-cross Abstract: LLM agents exhibit a consistent tendency to over-call, invoking tools even in situations where none is needed. On the When2Call benchmark, si...
arXiv:2606.16193v2 Announce Type: replace-cross Abstract: Multimodal Large Language Models (MLLMs) have demonstrated strong performance on vision-language tasks, yet their internal visual representat...
arXiv:2610.06814v2 Announce Type: replace-cross Abstract: World models learn to predict how their environment will evolve, making them an important foundation for general-purpose robotic control. Yet...
The study investigates whether particular attention heads and individual neurons within those heads in language models are responsible for detecting network infrastructure information—specifically hostnames paired with IP addresses. Using causal ablation and selective testing across five models from three architecture families, the authors find that a small subset of heads reliably identifies such information with near-perfect accuracy. However, the extent to which this responsibility is concentrated in a single neuron varies by model; in some cases a single neuron suffices, while in others the signal is distributed across the head. The findings generalize to an independent reverse‑DNS dataset, though single‑neuron detectors are less robust.
arXiv:2610.07083v1 Announce Type: cross Abstract: Driver fatigue poses a significant challenge to railway safety, with traditional systems like the dead-man switch offering limited and basic alertnes...
The paper introduces a Trust Layer for Agent Evaluation, a post‑hoc framework that appends a belief flag to each benchmark score. It checks four properties: alignment with the benchmark’s grading logic, traceable computation of a passing answer, consistency between the agent’s claim and actual behavior, and stability across repeated runs. Applied to five agent configurations on 108 tasks, the study finds that only 22.6% of recorded passes satisfy all checks, revealing widespread issues such as untraceable computation, false completion claims, and unstable results.
Learn2Play Bench is a new benchmark that tests how well large language model agents learn from experience in unfamiliar, text‑based games with novel or counterintuitive rules. The benchmark provides reproducible feedback, automatic scoring, and varied game instances to evaluate learning across repeated attempts and transfer to new situations. Findings show that retaining full action records aids learning, human players outperform agents, and the choice of harness significantly impacts performance and inference cost.
arXiv:2610.08662v1 Announce Type: new Abstract: As coding agents increasingly undertake real-world work autonomously, judging whether their risk treatments are warranted has become important. Existin...
The paper investigates whether frozen video‑language models inherently encode a signal indicating whether sufficient evidence has been observed to answer a question. By training linear probes on seven byte‑identical models, the authors demonstrate that these models contain a readable evidence‑readiness signal with AUROC ranging from 0.733 to 0.905, even when the probe is trained without any footage from the benchmark family. The signal is question‑conditioned, remains robust when the model answers incorrectly, and outperforms traditional uncertainty estimators; it can be leveraged as a Readiness Gating policy that improves answer accuracy by up to 9.75 percentage points without extra computational cost.
PolarScale is a new benchmark that explicitly requires models to predict the radiometric scale needed for full Stokes reconstruction from a per-scene normalized total‑intensity image. It evaluates models on normalized Stokes components, AoLP/DoLP/DoCP, and a per‑scene scale, using metrics that include angular, self‑consistency, and physical‑bound checks. Across seven restoration‑based and generative backbones, the best restoration models achieve a 3.6‑4.3% mean relative error in scale estimation, outperforming a constant‑scale control and maintaining physical plausibility.
arXiv:2610.08621v1 Announce Type: new Abstract: Recent game design agents have made substantial progress in generating playable games. However, program correctness does not ensure an enjoyable experi...
arXiv:2610.08402v1 Announce Type: new Abstract: Multi-turn LLM agents often receive sparse task feedback across several interactions, while generating each response token by token. This creates two r...
arXiv:2610.07324v1 Announce Type: cross Abstract: Time series foundation models (TSFMs) are trained on large collections of time series datasets that span various morphologies and domains. This setti...
The paper introduces Hierarchical Continuous Diffusion Language Models (H-CDLMs), a framework that jointly diffuses multiple token modalities—individual tokens and coarser clusters of token embeddings—to enhance continuous diffusion language models. Applied to the CoBit architecture, the resulting H-CoBit achieves significant empirical gains, improving MAUVE scores and achieving lower generative perplexity on LM1B and OWT, while also outperforming prior continuous diffusion models on GSM8K. The approach generalizes to other continuous generative paradigms, as shown by consistent improvements when applied to the flow matching model FLM.
Scribble-based interaction provides a lightweight and intuitive way for users to specify image editing intents in interactive editing tools. However, current image editing models based on VLMs or LLMs...
Urban diagnosis integrates heterogeneous observations to identify urban problems, localize affected areas, and investigate contributing factors, informing evidence-based urban planning and management....
Blind Source Separation(BSS) is a fundamental problem in signal processing, aiming to separate multiple source signals from their mixtures without prior knowledge of the sources or the mixing process....
Flow and diffusion models suffer from slow inference due to computationally expensive numerical integration. Distillation provides a promising way for a student model to learn from a teacher's dynamic...