Search: “concepts”

World Action Models: The Next Frontier in Embodied AI

arXiv:2605.12090v1 Announce Type: cross Abstract: Vision-Language-Action (VLA) models have achieved strong semantic generalization for embodied policy learning, yet they learn reactive observation-to-

safetyarxiv-cs-cl

13 May 2026

WorldComp2D: Spatio-semantic Representations of Object Identity and Location from Local Views

DGX agent

arXiv:2605.11743v1 Announce Type: new Abstract: Learning latent representations that capture both semantic and spatial information is central to efficient spatio-semantic reasoning. However, many exis

researcharxiv-cs-cv

13 May 2026

A Single Neuron Is Sufficient to Bypass Safety Alignment in Large Language Models

arXiv:2605.08513v1 Announce Type: cross Abstract: Safety alignment in language models operates through two mechanistically distinct systems: refusal neurons that gate whether harmful knowledge is expr

Autonomous FAIR Digital Objects: From Passive Assertions to Active Knowledge

arXiv:2605.10370v1 Announce Type: new Abstract: Scientific knowledge on the Web is published as passive assertions and cannot decide when to validate evidence, reconcile contradictions, or update conf

BEA-GS: BEyond RAdiance Supervision in 3DGS for Precise Object Extraction

arXiv:2605.09662v1 Announce Type: new Abstract: Most Gaussian Splatting techniques that provide a 3D semantic representation of the scene do not optimize the underlying 3D geometry, making object-leve

researcharxiv-cs-cv

Applications

Behavior-Centric Extraction of Scenarios from Highway Traffic Data and their Domain-Knowledge-Guided Clustering using CVQ-VAE

arXiv:2603.16964v2 Announce Type: replace Abstract: Approval of ADS depends on evaluating its behavior within representative real-world traffic scenarios. A common way to obtain such scenarios is to e

applicationsarxiv-cs-cv

Can You Keep a Secret? Involuntary Information Leakage in Language Model Writing

arXiv:2605.10794v1 Announce Type: cross Abstract: Language models are deployed in settings that require compartmentalization: system prompts should not be disclosed, chain-of-thought reasoning is hidd

CLEF: EEG Foundation Model for Learning Clinical Semantics

arXiv:2605.10817v1 Announce Type: new Abstract: Clinical EEG interpretation requires reasoning over full EEG sessions and integrating signal patterns with clinical context. Existing EEG foundation mod

CodeClinic: Evaluating Automation of Coding Skills for Clinical Reasoning Agents

arXiv:2605.09675v1 Announce Type: new Abstract: Clinical reasoning agents based on large language models (LLMs) aim to automate tasks such as intensive care unit (ICU) monitoring and patient state tra

Applications

Count Anything at Any Granularity

arXiv:2605.10887v1 Announce Type: new Abstract: Open-world object counting remains brittle: despite rapid advances in vision-language models (VLMs), reliably counting the objects a user intends is far

applicationsarxiv-cs-cv

Deep Dreams Are Made of This: Visualizing Monosemantic Features in Diffusion Models

arXiv:2605.08218v1 Announce Type: cross Abstract: This paper proposes latent visualization by optimization (LVO), a mechanistic interpretability technique that extends feature visualization by optimiz

researcharxiv-cs-cv

Diffusion Models are Evolutionary Algorithms

model-releasesarxiv-cs-lg

arXiv:2410.02543v3 Announce Type: replace-cross Abstract: In a convergence of machine learning and biology, we reveal that diffusion models are evolutionary algorithms. By considering evolution as a d

Explaining Graph Neural Networks for Node Similarity on Graphs

arXiv:2407.07639v2 Announce Type: replace-cross Abstract: Similarity search is a fundamental task for exploiting information in various applications dealing with graph data, such as citation networks

Exploring the AI Obedience: Why is Generating a Pure Color Image Harder than CyberPunk?

arXiv:2603.00166v2 Announce Type: replace-cross Abstract: Recent advances in generative AI have shown human-level performance in complex content creation. However, we identify a 'Paradox of Simplicity

Local Ai

fmxcoders: Factorized Masked Crosscoders for Cross-Layer Feature Discovery

arXiv:2605.09438v1 Announce Type: new Abstract: Many features in pretrained Transformers span multiple layers: they emerge through stages of inference, persist in the residual stream, or are built joi

local-aiarxiv-cs-lg

Functional Subspace, where language models can use vector algebra to solve problems

arXiv:2602.01687v2 Announce Type: replace-cross Abstract: Large language models (LLMs) were invented for natural language tasks such as translation, but they have proved that they can perform highly c

Generative Giants, Retrieval Weaklings: Why do Multimodal Large Language Models Fail at Multimodal Retrieval?

arXiv:2512.19115v2 Announce Type: replace Abstract: Despite the remarkable success of multimodal large language models (MLLMs) in generative tasks, we observe that they exhibit a counterintuitive defi

researcharxiv-cs-cv

GUARD: Guideline Upholding Test through Adaptive Role-play and Jailbreak Diagnostics for LLMs

arXiv:2508.20325v3 Announce Type: replace-cross Abstract: As Large Language Models (LLMs) become increasingly integral to various domains, their potential to generate harmful responses has prompted si

In-Context Fixation: When Demonstrated Labels Override Semantics in Few-Shot Classification

arXiv:2605.08295v1 Announce Type: cross Abstract: While random demonstration labels barely hurt in-context learning (Min et al., 2022), we show that homogeneous labels--even semantically valid ones--c

Interpretable Coreference Resolution Evaluation Using Explicit Semantics

arXiv:2605.10627v1 Announce Type: cross Abstract: Coreference resolution is typically evaluated using aggregate statistical metrics such as CoNLL-F1, which measure structural overlap between predicted

Language-Conditioned Visual Grounding with CLIP Multilingual

arXiv:2605.09060v1 Announce Type: new Abstract: Multilingual vision-language models exhibit systematic performance gaps across languages, but the mechanism remains ambiguous: cross-language divergence

researcharxiv-cs-cl

Learn the system

This article likely provides guidance on understanding how AI systems work, their underlying mechanisms, and best practices for effectively using or interacting with AI tools. Based on Ben's Bites' fo

tutorialsben-s-bites

Lecture Notes on Statistical Physics and Neural Networks

arXiv:2605.06394v1 Announce Type: cross Abstract: These lecture notes introduce some topics of classical statistical physics, particularly those that are relevant for neural networks and deep learning

LPT: Less-overfitting Prompt Tuning for Vision-Language Model

arXiv:2410.10247v3 Announce Type: replace-cross Abstract: Vision-language models (VLMs) have demonstrated exceptional generalization capabilities for downstream tasks. Due to its efficiency, prompt le

Applications

Medical Model Synthesis Architectures: A Case Study

arXiv:2605.09716v1 Announce Type: new Abstract: Medicine is rife with high-stakes uncertainty. Doctors routinely make clinical judgments and decisions that juggle many fundamental unknowns, like predi

applicationsarxiv-cs-ai

Mental Health AI Safety Claims Must Preserve Temporal Evidence

arXiv:2605.08827v1 Announce Type: new Abstract: The safety of mental health AI is often judged at the wrong temporal scale. Current evaluations typically score isolated responses, endpoint outcomes, o

MUR: Momentum Uncertainty guided Reasoning

arXiv:2507.14958v2 Announce Type: replace Abstract: Current models have achieved impressive performance on reasoning-intensive tasks, yet optimizing their reasoning efficiency remains an open challeng

tutorialsarxiv-cs-cl

Political Plasticity: An Analysis of Ideological Adaptability in Large Language Models

arXiv:2605.08415v1 Announce Type: new Abstract: Since the advent of Large Language Models (LLMs), a significant area of research has focused on their intrinsic biases, particularly in political discou

Reasoning Trajectories for Socratic Debugging of Student Code: From Misconceptions to Contradictions and Updated Beliefs

arXiv:2511.00371v2 Announce Type: replace Abstract: In Socratic debugging, instructors guide students towards identifying and fixing a bug on their own, instead of providing the bug fix directly. Most

tutorialsarxiv-cs-cl

Sens-VisualNews: A Benchmark Dataset for Sensational Image Detection

model-releasesarxiv-cs-cv

arXiv:2605.10394v1 Announce Type: new Abstract: The detection of sensational content in media items can be a critical filtering mechanism for identifying check-worthy content and flagging potential di

Threat Modelling using Domain-Adapted Language Models: Empirical Evaluation and Insights

arXiv:2605.10808v1 Announce Type: cross Abstract: Large Language Models(LLMs) are increasingly explored for cybersecurity applications such as vulnerability detection. In the domain of threat modellin

Time-Warping Recurrent Neural Networks for Transfer Learning

arXiv:2604.02474v2 Announce Type: replace Abstract: Dynamical systems describe how a physical system evolves over time. Physical processes can evolve faster or slower in different environmental condit

TINS: Test-time ID-prototype-separated Negative Semantics Learning for OOD Detection

model-releasesarxiv-cs-cv

arXiv:2605.10756v1 Announce Type: new Abstract: Vision-language models enable OOD detection by comparing image alignment with ID labels and negative semantics. Existing negative-label-based methods ma

Toward an Engineering of Science: Rebalancing Generation and Verification in the Age of AI

arXiv:2605.10425v1 Announce Type: cross Abstract: AI systems can now cheaply generate plausible scientific artifacts such as papers, reviews, and surveys. This creates a risk of epistemic pollution in

TrajPrism: A Multi-Task Benchmark for Language-Grounded Urban Trajectory Understanding

arXiv:2605.10782v1 Announce Type: new Abstract: Urban mobility is naturally expressed both as trajectories in space and as natural-language descriptions of travel intent, constraints, and preferences.

Unifying Scientific Communication: Fine-Grained Correspondence Across Scientific Media

model-releasesarxiv-cs-cv

arXiv:2605.05831v2 Announce Type: replace Abstract: The communication of scientific knowledge has become increasingly multimodal, spanning text, visuals, and speech through materials such as research

Your Simulation Runs but Solves the Wrong Physics: PDE-Grounded Intent Verification for LLM-Generated Multiphysics Simulation Code

arXiv:2605.09360v1 Announce Type: cross Abstract: Execution-based evaluation of LLM-generated code implicitly treats successful execution as a proxy for correctness. In scientific simulation, this pro

A Generalized Singular Value Theory for Neural Networks

arXiv:2605.06938v1 Announce Type: cross Abstract: Building on the abstract Generalized Singular Value Decomposition (GSVD) theory of Brown et al. [2025], we prove that most modern neural architectures

Active teacher selection for reward learning

arXiv:2310.15288v3 Announce Type: replace Abstract: Reward learning techniques enable machine learning systems to learn objectives from human feedback. A core limitation of these systems is their assu

tutorialsarxiv-cs-ai

An abstract effective convergence theorem for stochastic processes, with applications to stochastic approximation

arXiv:2504.12922v3 Announce Type: replace-cross Abstract: We provide a general theorem on the asymptotic behavior of stochastic processes that conform to a relaxed supermartingale condition. The disti

Cognitive Agent Compilation for Explicit Problem Solver Modeling

arXiv:2605.07040v1 Announce Type: cross Abstract: Large language models (LLMs) are widely used for tutoring, feedback generation, and content creation, but their broad pretraining makes them hard to c

Curvature Beyond Positivity: Greedy Guarantees for Arbitrary Submodular Functions

model-releasesarxiv-cs-lg

arXiv:2605.07902v1 Announce Type: new Abstract: Submodular functions -- functions exhibiting diminishing returns -- are central to machine learning. When the objective is monotone and non-negative, th

Demystifying Lipschitz verification: positive matrices, negative results

arXiv:2603.28113v2 Announce Type: replace Abstract: The global Lipschitz constant of a neural network is related to robustness and generalization, yet unlike in many classical models, it is not plainl

Local Ai

Emergent Manifold Separability during Reasoning in Large Language Models

arXiv:2602.20338v2 Announce Type: replace Abstract: Chain-of-Thought (CoT) prompting significantly improves reasoning in Large Language Models, yet the temporal dynamics of the underlying representati

local-aiarxiv-cs-lg

Equivalence of Coarse and Fine-Grained Models for Learning with Distribution Shift

arXiv:2605.07005v1 Announce Type: cross Abstract: Recent work on provably efficient algorithms for learning with distribution shift has focused on two models: PQ learning (Goldwasser et al. (2020)) an

From Standalone LLMs to Integrated Intelligence: A Survey of Compound Al Systems

arXiv:2506.04565v2 Announce Type: replace-cross Abstract: Compound AI Systems (CAIS) are an emerging paradigm that integrates large language models (LLMs) with external components, including retriever

researcharxiv-cs-cl

Identifiability Challenges in Sparse Linear Ordinary Differential Equations

arXiv:2506.09816v3 Announce Type: replace Abstract: Dynamical systems modeling is a core pillar of scientific inquiry across natural and life sciences. Increasingly, dynamical system models are learne

In-Context Credit Assignment via the Core

arXiv:2605.06920v1 Announce Type: cross Abstract: We propose incentive-aligned mechanisms for in-context credit assignment: the task of assigning credit for AI-generated content (e.g. code, news artic