Search: “safety”

Diagnosing Training Inference Mismatch in LLM Reinforcement Learning

arXiv:2605.14220v1 Announce Type: cross Abstract: Modern LLM RL systems separate rollout generation from policy optimization. These two stages are expected to produce token probabilities that match ex

did you know that Queen Elizabeth II wrote a Python graduate textbook?

DGX agent

did you know that Queen Elizabeth II wrote a Python graduate textbook? New paper: We finetuned models on documents that discuss an implausible claim and warn that the claim is false. Models ended up b

safetygary-marcus--x

15 May 2026

DiffusionOPD: A Unified Perspective of On-Policy Distillation in Diffusion Models

arXiv:2605.15055v1 Announce Type: cross Abstract: Reinforcement learning has emerged as a powerful tool for improving diffusion-based text-to-image models, but existing methods are largely limited to

Dimension-Level Intent Fidelity Evaluation for Large Language Models: Evidence from Structured Prompt Ablation

arXiv:2605.14517v1 Announce Type: cross Abstract: Holistic evaluation scores capture overall output quality but do not distinguish whether a model reproduced the structural form of a user's request fr

Distance-Matrix Wasserstein Statistics for Scalable Gromov--Wasserstein Learning

arXiv:2605.14981v1 Announce Type: new Abstract: Gromov--Wasserstein (GW) distances compare graphs, shapes, and point clouds through internal distances, without requiring a common coordinate system. Th

Distribution Corrected Offline Data Distillation for Large Language Models

arXiv:2605.14071v1 Announce Type: new Abstract: Distilling reasoning traces from strong large language models into smaller ones is a promising route to improve intelligence in resource-constrained set

Distributions as Actions: A Unified Framework for Diverse Action Spaces

arXiv:2506.16608v3 Announce Type: replace-cross Abstract: We introduce a novel reinforcement learning (RL) framework that treats parameterized action distributions as actions, redefining the boundary

Do Language Models Align with Brains? Prediction Scores Are Not Enough

arXiv:2605.14025v1 Announce Type: cross Abstract: Brain-language model comparisons often interpret neural prediction scores as evidence that model representations capture brain-relevant language compu

DSSP: Diffusion State Space Policy with Full-History Encoding

arXiv:2605.14598v1 Announce Type: new Abstract: Diffusion-based imitation learning has shown strong promise for robot manipulation. However, most existing policies condition only on the current observ

safetyarxiv-cs-ro

Dual Hierarchical Dialogue Policy Learning for Legal Inquisitive Conversational Agents

arXiv:2605.14057v1 Announce Type: new Abstract: Most existing dialogue systems are user-driven, primarily designed to fulfill user requests. However, in many critical real-world scenarios, a conversat

Dynamic Mixed-Precision Routing for Efficient Multi-step LLM Interaction

arXiv:2602.02711v2 Announce Type: replace Abstract: Large language models (LLMs) achieve strong performance in long-horizon decision-making tasks through multi-step interaction and reasoning at test t

Efficient Generative Retrieval for E-commerce Search with Semantic Cluster IDs and Expert-Guided RL

arXiv:2605.14434v1 Announce Type: cross Abstract: Generative retrieval offers a promising alternative by unifying the fragmented multi-stage retrieval process into a single end-to-end model. However,

EponaV2: Driving World Model with Comprehensive Future Reasoning

arXiv:2605.14696v1 Announce Type: new Abstract: Data scaling plays a pivotal role in the pursuit of general intelligence. However, the prevailing perception-planning paradigm in autonomous driving rel

Every Subtlety Counts: Fine-grained Person Independence Micro-Action Recognition via Distributionally Robust Optimization

arXiv:2509.21261v3 Announce Type: replace Abstract: Micro-action Recognition is vital for psychological assessment and human-computer interaction. However, existing methods often fail in real-world sc

Evo-Depth: A Lightweight Depth-Enhanced Vision-Language-Action Model

arXiv:2605.14950v1 Announce Type: new Abstract: Vision-Language-Action models have emerged as a promising paradigm for robotic manipulation by unifying perception, language grounding, and action gener

Evolving Layer-Specific Scalar Functions for Hardware-Aware Transformer Adaptation

arXiv:2605.14047v1 Announce Type: new Abstract: Vision Transformers (ViTs) achieve state-of-the-art performance on challenging vision tasks, but their deployment on edge devices is severely hindered b

FlowSteer: Towards Agents Designing Agentic Workflows via Reinforced Progressive Canvas Editing

arXiv:2602.01664v4 Announce Type: replace Abstract: In recent years, agentic workflows have been widely applied to solve complex human tasks. However, existing workflow construction still faces key ch

From Ranking to Reasoning: Explainable Web API Recommendation via Semantic Reasoning

arXiv:2511.05820v2 Announce Type: replace-cross Abstract: The rapid growth of Web APIs has made automated Web API recommendation essential for efficient mashup development. However, existing approache

@GaryMarcus If token prediction is 'generating thought itself,' then my phone's autocomplete is a philosopher.

Gary Marcus critiques the claim that token prediction in large language models constitutes genuine thought or reasoning, using the analogy of smartphone autocomplete to illustrate that predictive text

safetygary-marcus--x

Generative Deep Learning for Computational Destaining and Restaining of Unregistered Digital Pathology Images

arXiv:2605.14251v1 Announce Type: new Abstract: Conditional generative adversarial networks (cGANs) have enabled high-fidelity computational staining and destaining of hematoxylin and eosin (H&E) in d

Google confirms it's testing a new storage policy after some users reported that new Gmail accounts get only 5GB of free storage unless they add a phone number (Akshay Gangwar/Android Authority)

Akshay Gangwar / Android Authority: Google confirms it's testing a new storage policy after some users reported that new Gmail accounts get only 5GB of free storage unless they add a phone number — Up

safetytechmeme

Google updates its spam rules to include attempts to ‘manipulate’ AI

Google updated its spam policy to mark attempts to 'manipulate' its AI model in search results as spam, including results in AI Overview or AI Mode in Search, as Search Engine Land reports: 'In the co

safetythe-verge-ai

Grokking Finite-Dimensional Algebra

arXiv:2602.19533v2 Announce Type: replace-cross Abstract: This paper investigates the grokking phenomenon, which refers to the sudden transition from a long memorization to generalization observed dur

Hand-in-the-Loop: Improving Dexterous VLA via Seamless Interventional Correction

arXiv:2605.15157v1 Announce Type: cross Abstract: Vision-Language-Action (VLA) models are prone to compounding errors in dexterous manipulation, where high-dimensional action spaces and contact-rich d

HeatKV: Head-tuned KV-cache Compression for Visual Autoregressive Modeling

arXiv:2605.14877v1 Announce Type: new Abstract: Visual Autoregressive (VAR) models have recently demonstrated impressive image generation quality while maintaining low latency. However, they suffer fr

Hierarchical Image Tokenization for Multi-Scale Image Super Resolution

arXiv:2605.14891v1 Announce Type: new Abstract: We introduce a multi-scale Image Super Resolution (ISR) method building on recent advances in Visual Auto-Regressive (VAR) modeling. VAR models break im

Hyperbolic Graph Neural Networks Under the Microscope: The Role of Geometry-Task Alignment

arXiv:2602.01828v2 Announce Type: replace Abstract: Many complex networks exhibit hierarchical, tree-like structures, making hyperbolic space a natural candidate wherein to learn representations of th

ICED: Concept-level Machine Unlearning via Interpretable Concept Decomposition

arXiv:2605.14309v1 Announce Type: cross Abstract: Machine unlearning in Vision-Language Models (VLMs) is typically performed at the image or instance level, making it difficult to precisely remove tar

Identifying Culprits Through Deep Deterministic Policy Gradient Deep Learning Investigation

arXiv:2605.14774v1 Announce Type: new Abstract: In the world of AI and advanced technologies investigation aspects identification of a crime or criminal plays a major problem. In this research we focu

Ideology Prediction of German Political Texts

arXiv:2605.14352v1 Announce Type: new Abstract: Elections represent a crucial milestone in a nation's ongoing development. To better understand the political rhetoric from various movements, ranging f

'In the past nine months, the United States has produced more AI legislation than in the prior decade,' write @JeffSonnenfeld, @GaryMarcus, …

'In the past nine months, the United States has produced more AI legislation than in the prior decade,' write @JeffSonnenfeld, @GaryMarcus, and Stephen Henriques in a commentary piece for Fortune. 'No

safetygary-marcus--x

InfoSFT: Learn More and Forget Less with Information-Aware Token Weighting

arXiv:2605.14967v1 Announce Type: new Abstract: Supervised fine-tuning (SFT) provides the standard approach for teaching LLMs new behaviors from offline expert demonstrations. However, standard SFT un

Know When To Fold 'Em: Token-Efficient LLM Synthetic Data Generation via Multi-Stage In-Flight Rejection

arXiv:2605.14062v1 Announce Type: new Abstract: While synthetic data generation with large language models (LLMs) is widely used in post-training pipelines, existing approaches typically generate full

KVPO: ODE-Native GRPO for Autoregressive Video Alignment via KV Semantic Exploration

arXiv:2605.14278v1 Announce Type: new Abstract: Aligning streaming autoregressive (AR) video generators with human preferences is challenging. Existing reinforcement learning methods predominantly rel

Language Generation as Optimal Control: Closed-Loop Diffusion in Latent Control Space

arXiv:2605.14531v1 Announce Type: new Abstract: This work reformulates language generation as a stochastic optimal control problem, providing a unified theoretical perspective to analyze autoregressiv

LATERN: Test-Time Context-Aware Explainable Video Anomaly Detection

arXiv:2605.15054v1 Announce Type: new Abstract: Vision-language models (VLMs) have recently emerged as a promising paradigm for video anomaly detection (VAD) due to their strong visual reasoning abili

Learning from Failures: Correction-Oriented Policy Optimization with Verifiable Rewards

arXiv:2605.14539v1 Announce Type: new Abstract: Reinforcement Learning with Verifiable Rewards (RLVR) has emerged as an effective paradigm for improving the reasoning capabilities of large language mo

Learning from Language Feedback via Variational Policy Distillation

arXiv:2605.15113v1 Announce Type: new Abstract: Reinforcement learning from verifiable rewards (RLVR) suffers from sparse outcome signals, creating severe exploration bottlenecks on complex reasoning

Learning to Build the Environment: Self-Evolving Reasoning RL via Verifiable Environment Synthesis

arXiv:2605.14392v1 Announce Type: new Abstract: We pursue a vision for self-improving language models in which the model does not merely generate problems or traces to imitate, but constructs the envi

Let Robots Feel Your Touch: Visuo-Tactile Cortical Alignment for Embodied Mirror Resonance

arXiv:2605.14571v1 Announce Type: cross Abstract: Observing touch on another's body can elicit corresponding tactile sensations in the observer, a phenomenon termed mirror touch that supports empathy

Logging Policy Design for Off-Policy Evaluation

arXiv:2605.15108v1 Announce Type: cross Abstract: Off-policy evaluation (OPE) estimates the value of a target treatment policy (e.g., a recommender system) using data collected by a different logging

LPH-VTON: Resolving the Structure-Texture Dilemma of Virtual Try-On via Latent Process Handover

arXiv:2605.14874v1 Announce Type: new Abstract: Virtual Try-On (VTON) aims to synthesize photorealistic images of garments precisely aligned with a person's body and pose. Current diffusion-based meth

Mechanical Enforcement for LLM Governance:Evidence of Governance-Task Decoupling in Financial Decision Systems

arXiv:2605.14744v1 Announce Type: cross Abstract: Large language models in regulated financial workflows are governed by natural-language policies that the same model interprets, creating a principal-

Medical Report Generation: A Hierarchical Task Structure-Based Cross-Modal Causal Intervention Framework

arXiv:2511.02271v2 Announce Type: replace Abstract: Medical Report Generation (MRG) is a key part of modern medical diagnostics, as it automatically generates reports from radiological images to reduc

Mitigating Mask Prior Drift and Positional Attention Collapse in Large Diffusion Vision-Language Models

arXiv:2605.14530v1 Announce Type: new Abstract: Large diffusion vision-language models (LDVLMs) have recently emerged as a promising alternative to autoregressive models, enabling parallel decoding fo

MoRe: Modular Representations for Principled Continual Representation Learning on Squantial Data

arXiv:2605.14364v1 Announce Type: new Abstract: Continual learning requires models to adapt to new data while preserving previously acquired knowledge. At its core, this challenge can be viewed as pri

Multi-Dimensional Model Integrity and Responsibility Assessment Index and Scoring Framework

arXiv:2605.14550v1 Announce Type: new Abstract: Artificial intelligence in high-stakes tabular domains cannot be evaluated by predictive performance alone, yet current practice still assesses explaina

Multi-scale Coarse-to-fine Modeling for Test-time Human Motion Control

arXiv:2605.14935v1 Announce Type: new Abstract: We present MSCoT, a multi-scale, coarse-to-fine model for test-time human motion synthesis and control. Unlike recent approaches that rely on multiple i