Search: “safety”

E^2-OPSD: Taming Entropy Overshoot in On-Policy Self-Distillation

arXiv:2610.05048v1 Announce Type: new Abstract: On-policy self-distillation (OPSD) provides dense token-level supervision without a second model: one network acts as teacher with the reference solutio

Human

Content type

All Blog X Post Paper YouTube Reddit GitHub

Clear filters

EasyClassifier: Honest, Reproducible Machine-Learning Classification for Researchers Who Do Not Program

DGX agent

arXiv:2610.04758v1 Announce Type: new Abstract: Machine-learning classification is now used across medicine, the social sciences, economics, education and engineering, very often by researchers who do

safetyarxiv-cs-lg

6 Oct 2026

EduAlign: Aligning Educational LLMs for Helpfulness, Personalization, and Creativity

arXiv:2507.20335v2 Announce Type: replace Abstract: Educational language models should pursue three complementary objectives: helpfulness through responsible guidance, personalization to learner needs

EgoSteer: An Open-Source Full-Stack System Towards Steerable Dexterous Manipulation from Egocentric Videos

arXiv:2607.09701v2 Announce Type: replace Abstract: The enduring vision of general-purpose robots serving humanity hinges fundamentally on policy steerability. However, prevailing paradigms of learnin

Encoded but Not in Control: Revealing the Grounding Gap in Vision-Language Robot Policies

arXiv:2610.06235v1 Announce Type: cross Abstract: Instruction following is central to language-conditioned robot policies: language should determine what to do when the same scene permits multiple val

Erased, Rerouted, or Rescaled? Post-Training and the Causal Quotient of a Language Model's Belief State

arXiv:2610.05292v1 Announce Type: new Abstract: What happens to information a pretrained model already encodes when post-training no longer rewards using it? The common language of representation comp

Evolving in Thought Space: Training a Small Model at Test Time Unlocks Better Discoveries

arXiv:2610.06269v1 Announce Type: cross Abstract: Open-ended scientific discovery often requires repeatedly proposing and evaluating candidate solutions. LLM-based systems can support this process by

Evolving LLM-Generated Features for Interpretable Classification

arXiv:2610.03951v1 Announce Type: new Abstract: Large language models (LLMs) are increasingly used as classifiers, yet they operate as opaque systems whose decisions are difficult to interpret, which

ExpertMuon-Compass: Alignment-Guided Step Sizes for Mixture-of-Experts Training

arXiv:2610.04140v1 Announce Type: new Abstract: Mixture-of-experts (MoE) language models send each token to a few experts, so each expert is trained on a different part of the data, and this part chan

ExStereo: Lifting 2D Vision-Language-Action Models to 3D with Explicit Stereo Representations

arXiv:2610.04805v1 Announce Type: cross Abstract: Three-dimensional perception is critical for robotic manipulation, particularly for high-precision tasks, as recovering metric depth and precise 3D ob

FairProp: Fair Node Representation Learning via Differentiable Propagation Layers

arXiv:2610.06484v1 Announce Type: new Abstract: Graph neural networks (GNNs) are the standard tool for node representation learning and are increasingly used in high-stakes settings. Their message-pas

Fast Last-Iterate Convergence in Zero-Sum Markov Games with Bandit Feedback

arXiv:2610.05968v1 Announce Type: new Abstract: We study last-iterate convergence in unknown two-player zero-sum discounted Markov games with bandit feedback. The players learn independently along a s

Fine-Grained Caching for Diffusion Transformers with Few Calibration Conditions

arXiv:2512.05134v2 Announce Type: replace Abstract: Diffusion transformers require repeated denoiser evaluations, making image and video generation computationally expensive. We propose a training-fre

Finite-Sample Distribution Theory and Efficient Large-Scale Inference for Online Quantile Regression

arXiv:2610.05869v1 Announce Type: cross Abstract: This paper studies online quantile regression for large-scale and streaming data using Stochastic SubGradient Descent (SSGD) with constant learning ra

Flash-OPD: Fast On-Policy Distillation

arXiv:2610.06105v1 Announce Type: new Abstract: On-policy distillation (OPD) provides dense teacher supervision on student-generated trajectories, but generating and evaluating long rollouts incurs su

FLEX-WAM: Flexible Block-Causal World-Action Models for Long-Horizon Imagination and Planning

arXiv:2610.05483v1 Announce Type: new Abstract: World--action models (WAMs) promise a unified model that predicts action-conditioned futures, generates feasible actions, and supports planning in imagi

Flip-Flop Consistency: Unsupervised Training for Robustness to Prompt Perturbations in LLMs

arXiv:2510.14242v2 Announce Type: replace-cross Abstract: Large Language Models (LLMs) often produce inconsistent answers when faced with different phrasings of the same prompt. In this paper, we prop

Flow Policies as Actions of Skill-Level World Models: Learned and Symbolic Abstractions for Long-Horizon Planning

arXiv:2610.04767v1 Announce Type: cross Abstract: Latent world models enable robots to plan by predicting the consequences of actions. Planning long tasks with control-rate actions requires many predi

FoleyGenEx: Unified Video-to-Audio Generation with Multi-Modal Control, Temporal Alignment, and Semantic Precision

arXiv:2606.14049v2 Announce Type: replace-cross Abstract: We present FoleyGenEx, a unified video-to-audio (VTA) framework integrating multi-modal control, frame-level temporal alignment, and fine-grai

Frame-Level Temporal Alignment for Human-to-Robot Visual Adaptation

arXiv:2610.04372v1 Announce Type: cross Abstract: Transferring visual representations pretrained on human videos to robot manipulation requires learning reliable correspondences between human and robo

Future Anchored Verification and Online Recovery for World Action Models

arXiv:2610.06280v1 Announce Type: new Abstract: World action models (WAMs) have emerged as a promising paradigm for robotic manipulation. They act by first predicting how a task should be performed an

GAMBIT: Learning to Plan Continuous Multi-Robot Trajectories

arXiv:2610.06290v1 Announce Type: cross Abstract: GAMBIT is an opening chess move in which a player sacrifices a piece, typically a pawn, to gain a positional advantage later in the game. Analogously,

Geometry-Aware Preference Optimization for Text-to-Image Diffusion Models

arXiv:2610.04980v1 Announce Type: new Abstract: Preference alignment has become a standard practice for text-to-image diffusion models. Direct Preference Optimization (DPO) simplifies this process by

GRAM: Correcting Frozen Time-Series Foundation Models via Graph-Retrieved Amplitude Memory

arXiv:2610.04827v1 Announce Type: new Abstract: Time-series foundation models (TSFMs) enable zero-shot forecasting through large-scale cross-domain pretraining, while retrieval augmentation further im

Groupwise Distortion Guarantees for Preference-Based Alignment

arXiv:2610.05450v1 Announce Type: new Abstract: Preference-based alignment methods such as reinforcement learning from human feedback (RLHF) and Nash learning from human feedback (NLHF) aggregate pair

GUIDE: Reinforcement Learning for Behavioral Action Support in Type 1 Diabetes

arXiv:2604.00385v2 Announce Type: replace Abstract: Type 1 diabetes (T1D) management requires continuous adjustment of insulin and lifestyle behaviors to maintain blood glucose within a safe target ra

Hierarchical Reinforcement Learning for Collision-Free Locomotion of an Underactuated Biped

arXiv:2610.05855v1 Announce Type: new Abstract: A bipedal robot cannot deviate from its path to avoid an obstacle without disturbing its balance, and this coupling is most severe on underactuated plat

Hierarchical Time-aware Bootstrapping for Off-Policy Subgoal Value Learning

arXiv:2610.05446v1 Announce Type: new Abstract: Off-policy hierarchical reinforcement learning must estimate the values of high-level decisions while the low-level policy changes. HIRO adapts replay d

How Should Teachers Be Prepared? RL on Student-Induced States for On-Policy Distillation

arXiv:2610.04950v1 Announce Type: new Abstract: On-policy distillation (OPD) improves the reasoning capabilities of small language models through token-level teacher supervision on student-generated t

Humanoid Rickshaw Pulling: Whole-Body Locomotion under Coupled Wheeled Loads

arXiv:2610.04238v1 Announce Type: cross Abstract: Humanoid robots could transport payloads substantially heavier than themselves by pulling passive wheeled vehicles instead of carrying the load. This

I-BFM: Reward-Conditioned Robust Humanoid Interaction via Unsupervised Reinforcement Learning

arXiv:2610.06129v1 Announce Type: new Abstract: Behavioral foundation models (BFMs) have recently shown that a single humanoid policy can support diverse whole-body control, but extending such general

ilde{O}(sqrt{T}) Regret and Polylogarithmic Constraint Violation for COCO

arXiv:2610.03983v1 Announce Type: new Abstract: We study constrained online convex optimization with adversarial convex losses and constraints (mathsf{COCO}). At each round (tin[T]), a learner selects

Incentive Alignment in Online Experimentation

arXiv:2610.05922v1 Announce Type: cross Abstract: Evaluating the causal effect of new features is a central goal for online platforms. While recent literature addresses limited testing traffic via cen

InterMimicGen: Scaling Humanoid Loco-Manipulation through Self-Evolving Motion Imitation

arXiv:2610.06850v1 Announce Type: cross Abstract: Captured human-object interactions provide rich supervision for humanoid loco-manipulation, but they are sparse, heterogeneous, and not directly execu

Investors pattern matching on old movies is going to lead to a recession. There is no network effect here and no reason to think that large …

Investors pattern matching on old movies is going to lead to a recession. There is no network effect here and no reason to think that large language models will be anything other than a brutal commodi

safetygary-marcus--x

Kandinsky 6.0 Video: Foundation Models for Synchronized Video and Audio Generation

arXiv:2610.05608v1 Announce Type: new Abstract: We present Kandinsky 6.0 Video, a family of foundation diffusion models for synchronized text-to-audio-video generation, comprising Kandinsky 6.0 Video

Learning Coordinated Visuomotor Box-Pushing from Solo Demonstrations

arXiv:2610.05677v1 Announce Type: new Abstract: Multi-robot imitation learning, particularly in settings where visuomotor policies are deployed in a communication-free, onboard decentralised style, re

Learning Pareto Stationary Fronts via Single-Pass Backpropagation

arXiv:2610.06397v1 Announce Type: new Abstract: We propose MOSEL (Multi-Objective Stackelberg Efficient Learning), a framework for a posteriori multi-objective optimization (MOO) in deep neural networ

Learning to Read the Contextual Tokens in Diffusion Transformers

arXiv:2610.06844v1 Announce Type: new Abstract: Multimodal Diffusion Transformers (MM-DiTs) jointly process visual and textual representations throughout generation. These models repeatedly update the

Learning What to Imitate: Entropy-Aware Distribution Mixing

arXiv:2610.06671v1 Announce Type: new Abstract: Small language models are often post-trained as students on reasoning traces from stronger teacher models to efficiently learn new skills. However, toke

Learning without Overwriting: A Theory of Self-Distillation and Supervised Fine-Tuning in Continual Reasoning

arXiv:2610.05200v1 Announce Type: new Abstract: On-policy self-distillation (OPSD) of large language models (LLMs) has demonstrated the ability to improve reasoning capabilities while preserving previ

LeAVJEPA: A Minimalist Architecture for Audio-Visual Self-Supervised Learning

arXiv:2610.06226v1 Announce Type: cross Abstract: Prior audio-visual self-supervised learning methods rely on mechanisms such as EMA target encoders, prediction heads, reconstruction decoders, and con

Lie Rarely, Lie Big: Stealthy Insider Attacks on LLM Robot Teams

arXiv:2610.04744v1 Announce Type: new Abstract: When a team of robots delegates planning and mutual trust to LLM agents, a single compromised robot can corrupt the shared outcome. We study this threat

Local2Mesh: Spatially Localized Contour-to-Mesh for Left Ventricular Reconstruction from Sparse 2D Cardiac MRI

arXiv:2610.06052v1 Announce Type: new Abstract: Three-dimensional (3D) left ventricular (LV) reconstruction from sparse cardiac magnetic resonance (CMR) imaging remains challenging due to inter-slice

Long-MDR: Long-Context Reinforcement Learning for Multimodal Deep-Research Agents

arXiv:2610.05195v1 Announce Type: new Abstract: The next generation of multimodal research agents must reason over long-lived research histories rather than short model completions. During a single ta

Loss-Invariant Projections as Passive Probes of Learned Representations

arXiv:2610.06195v1 Announce Type: cross Abstract: Learned feature representations in neural networks often contain structure beyond that directly used by the final task output. We study this structure

MAGEFormer: Learning Metric-Consistent Representations for Anisotropic CT Segmentation

arXiv:2610.04036v1 Announce Type: new Abstract: Vision Transformers (ViTs) have shown strong performance in volumetric segmentation, but their effectiveness on clinical CT is limited by an isotropic E

MambaVF: State Space Model for Efficient Video Fusion

arXiv:2602.06017v2 Announce Type: replace Abstract: Video fusion aims to integrate complementary information from multiple source videos while preserving temporal consistency. Effective modeling of te