Search: “safety”

Let me say this clearly: LLMs cannot feel emotions. Emotions are evolutionary mechanisms. They push us to avoid danger or approach what is b…

Let me say this clearly: LLMs cannot feel emotions. Emotions are evolutionary mechanisms. They push us to avoid danger or approach what is beneficial. We experience emotions because we are alive, and

safetygary-marcus--x

Content type

All Blog X Post Paper YouTube Reddit GitHub

Clear filters

Safety

🦤 LeWorldModel: Learning Physics from Pixels — Stable World Models with Just Two Losses World models: 1️⃣ DINO-WM: pretrained ViT encoder (…

DGX agent

🦤 LeWorldModel: Learning Physics from Pixels — Stable World Models with Just Two Losses World models: 1️⃣ DINO-WM: pretrained ViT encoder (from ImageNet) → features → predictor. But encoder is frozen,

safetyyann-lecun--x

22 Apr 2026

LiteParse: our open-source, layout-aware PDF parser for AI agents. The secret? Grid projection. Instead of heavy ML layout models or flat te…

LiteParse: our open-source, layout-aware PDF parser for AI agents. The secret? Grid projection. Instead of heavy ML layout models or flat text extraction, it projects text onto a monospace grid so ali

safetyjerry-liu--x

Location Not Found: Exposing Implicit Local and Global Biases in Multilingual LLMs

arXiv:2604.19292v1 Announce Type: cross Abstract: Multilingual large language models (LLMs) have minimized the fluency gap between languages. This advancement, however, exposes models to the risk of b

LogosKG: Hardware-Optimized Scalable and Interpretable Knowledge Graph Retrieval

arXiv:2604.18913v1 Announce Type: new Abstract: Knowledge graphs (KGs) are increasingly integrated with large language models (LLMs) to provide structured, verifiable reasoning. A core operation in th

Low-Rank Adaptation for Critic Learning in Off-Policy Reinforcement Learning

arXiv:2604.18978v1 Announce Type: cross Abstract: Scaling critic capacity is a promising direction for enhancing off-policy reinforcement learning (RL). However, larger critics are prone to overfittin

Lyapunov-Certified Direct Switching Theory for Q-Learning

arXiv:2604.19569v1 Announce Type: cross Abstract: Q-learning is one of the most fundamental algorithms in reinforcement learning. We analyze constant-stepsize Q-learning through a direct stochastic sw

M^{2}GRPO: Mamba-based Multi-Agent Group Relative Policy Optimization for Biomimetic Underwater Robots Pursuit

arXiv:2604.19404v1 Announce Type: cross Abstract: Traditional policy learning methods in cooperative pursuit face fundamental challenges in biomimetic underwater robots, where long-horizon decision ma

Machine individuality: Separating genuine idiosyncrasy from response bias in large language models

arXiv:2604.16755v2 Announce Type: replace Abstract: As large language models (LLMs) are increasingly integrated into daily life, in roles ranging from high-stakes decision support to companionship, un

MacroNav: Multi-Task Context Representation Learning Enables Efficient Navigation in Unknown Environments

arXiv:2511.04320v2 Announce Type: replace Abstract: Autonomous navigation in unknown environments requires multi-scale spatial understanding that captures geometric details, topological connectivity,

safetyarxiv-cs-ro

Mask World Model: Predicting What Matters for Robust Robot Policy Learning

arXiv:2604.19683v1 Announce Type: new Abstract: World models derived from large-scale video generative pre-training have emerged as a promising paradigm for generalist robot policy learning. However,

safetyarxiv-cs-ro

Mitigating Judgment Preference Bias in Large Language Models through Group-Based Polling

arXiv:2510.08145v2 Announce Type: replace Abstract: Large Language Models (LLMs) as automatic evaluators, commonly referred to as LLM-as-a-Judge, have also attracted growing attention. This approach p

Mitigating Long-Tail Bias via Prompt-Controlled Diffusion Augmentation

arXiv:2602.04749v2 Announce Type: replace Abstract: Long-tailed class imbalance remains a fundamental obstacle in semantic segmentation of high-resolution remote-sensing imagery, where dominant classe

Mixture of Predefined Experts: Maximizing Data Usage on Vertical Federated Learning

arXiv:2602.12708v2 Announce Type: replace Abstract: Vertical Federated Learning (VFL) has emerged as a critical paradigm for collaborative model training in privacy-sensitive domains such as finance a

MMControl: Unified Multi-Modal Control for Joint Audio-Video Generation

arXiv:2604.19679v1 Announce Type: new Abstract: Recent advances in Diffusion Transformers (DiTs) have enabled high-quality joint audio-video generation, producing videos with synchronized audio within

MOSA: Motion-Guided Semantic Alignment for Dynamic Scene Graph Generation

arXiv:2604.19631v1 Announce Type: new Abstract: Dynamic Scene Graph Generation (DSGG) aims to structurally model objects and their dynamic interactions in video sequences for high-level semantic under

MRS: Multi-Resolution Skills for HRL Agents

arXiv:2505.21410v2 Announce Type: replace Abstract: Hierarchical reinforcement learning (HRL) decomposes the policy into a manager and a worker, enabling long-horizon planning but introducing a perfor

Multi-Gait Learning for Humanoid Robots Using Reinforcement Learning with Selective Adversarial Motion Prior

arXiv:2604.19102v1 Announce Type: cross Abstract: Learning diverse locomotion skills for humanoid robots in a unified reinforcement learning framework remains challenging due to the conflicting requir

Multi-modal Reasoning with LLMs for Visual Semantic Arithmetic

arXiv:2604.19567v1 Announce Type: new Abstract: Reinforcement learning (RL) as post-training is crucial for enhancing the reasoning ability of large language models (LLMs) in coding and math. However,

Multi-Task Reinforcement Learning for Enhanced Multimodal LLM-as-a-Judge

arXiv:2603.11665v2 Announce Type: replace Abstract: Multimodal Large Language Models (MLLMs) have been widely adopted as MLLM-as-a-Judges due to their strong alignment with human judgment across vario

Multiclass Local Calibration with the Jensen-Shannon Distance

arXiv:2510.26566v2 Announce Type: replace-cross Abstract: Developing trustworthy Machine Learning (ML) models requires their predicted probabilities to be well-calibrated, meaning they should reflect

On the Generalizability of Foundation Models for Crop Type Mapping

arXiv:2409.09451v5 Announce Type: replace Abstract: Foundation models pre-trained using self-supervised learning have shown powerful transfer learning capabilities on various downstream tasks, includi

Online Learning of Whittle Indices for Restless Bandits with Non-Stationary Transition Kernels

arXiv:2506.18186v3 Announce Type: replace Abstract: The restless multi-armed bandit (RMAB) framework is a popular approach to solving resource allocation problems in networked systems. In this paper,

Personalized Benchmarking: Evaluating LLMs by Individual Preferences

arXiv:2604.18943v1 Announce Type: new Abstract: With the rise in capabilities of large language models (LLMs) and their deployment in real-world tasks, evaluating LLM alignment with human preferences

Persuasion with Large Language Models: A Survey of Empirical Evidence, Study Methodologies, and Ethical Implications

arXiv:2411.06837v2 Announce Type: replace Abstract: The rapid rise of Large Language Models (LLMs) has created new disruptive possibilities for persuasive communication, enabling fully-automated, pers

Phase-Aware Policy Learning for Skateboard Riding of Quadruped Robots via Feature-wise Linear Modulation

arXiv:2602.09370v2 Announce Type: replace Abstract: Skateboards offer a compact and efficient means of transportation as a type of personal mobility device. However, controlling them with legged robot

safetyarxiv-cs-ro

Policy Gradient Primal-Dual Method for Safe Reinforcement Learning from Human Feedback

arXiv:2604.19024v1 Announce Type: new Abstract: Safe Reinforcement Learning from Human Feedback (Safe RLHF) has recently achieved empirical success in developing helpful and harmless large language mo

Prioritizing the Best: Incentivizing Reliable Multimodal Reasoning by Rewarding Beyond Answer Correctness

arXiv:2604.18892v1 Announce Type: new Abstract: Reinforcement Learning with Verifiable Rewards (RLVR) improves multimodal reasoning by rewarding verifiable final answers. Yet answer-correct trajectori

Probing for Reading Times

arXiv:2604.18712v1 Announce Type: new Abstract: Probing has shown that language model representations encode rich linguistic information, but it remains unclear whether they also capture cognitive sig

Product-of-Experts Training Reduces Dataset Artifacts in Natural Language Inference

arXiv:2604.19069v1 Announce Type: cross Abstract: Neural NLI models overfit dataset artifacts instead of truly reasoning. A hypothesis-only model gets 57.7% in SNLI, showing strong spurious correlatio

Proposing Topic Models and Evaluation Frameworks for Analyzing Associations with External Outcomes: An Application to Leadership Analysis Using Large-Scale Corporate Review Data

arXiv:2604.18919v1 Announce Type: new Abstract: Analyzing topics extracted from text data in relation to external outcomes is important across fields such as computational social science and organizat

QTMRL: An Agent for Quantitative Trading Decision-Making Based on Multi-Indicator Guided Reinforcement Learning

arXiv:2508.20467v2 Announce Type: replace-cross Abstract: In the highly volatile and uncertain global financial markets, traditional quantitative trading models relying on statistical modeling or empi

Quantifying Data Similarity Using Cross Learning

arXiv:2510.10866v3 Announce Type: replace-cross Abstract: Measuring dataset similarity is fundamental in machine learning, particularly for transfer learning and domain adaptation. In the context of s

Reasoning-Aware AIGC Detection via Alignment and Reinforcement

arXiv:2604.19172v1 Announce Type: new Abstract: The rapid advancement and widespread adoption of Large Language Models (LLMs) have elevated the need for reliable AI-generated content (AIGC) detection,

Model Releases

ReflectMT: Internalizing Reflection for Efficient and High-Quality Machine Translation

model-releasesarxiv-cs-cl

arXiv:2604.19144v1 Announce Type: new Abstract: Recent years have witnessed growing interest in applying Large Reasoning Models (LRMs) to Machine Translation (MT). Existing approaches predominantly ad

Regulating Artificial Intimacy: From Locks and Blocks to Relational Accountability

arXiv:2604.18893v1 Announce Type: cross Abstract: A series of high-profile tragedies involving companion chatbots has triggered an unusually rapid regulatory response. Several jurisdictions, including

Reinforcement Learning Improves LLM Accuracy and Reasoning in Disease Classification from Radiology Reports

arXiv:2604.19060v1 Announce Type: new Abstract: Accurate disease classification from radiology reports is essential for many applications. While supervised fine-tuning (SFT) of lightweight LLMs improv

Model Releases

RepIt: Steering Language Models with Concept-Specific Refusal Vectors

model-releasesarxiv-cs-ai

arXiv:2509.13281v5 Announce Type: replace Abstract: Current safety evaluations of language models rely on benchmark-based assessments that may miss localized vulnerabilities. We present RepIt, a simpl

RESFL: An Uncertainty-Aware Framework for Responsible Federated Learning by Balancing Privacy, Fairness and Utility

arXiv:2503.16251v2 Announce Type: replace-cross Abstract: Federated Learning (FL) has gained prominence in machine learning applications across critical domains by enabling collaborative model trainin

RL-ABC: Reinforcement Learning for Accelerator Beamline Control

arXiv:2604.19146v1 Announce Type: new Abstract: Particle accelerator beamline optimization is a high-dimensional control problem traditionally requiring significant expert intervention. We present RLA

Semantic Needles in Document Haystacks: Sensitivity Testing of LLM-as-a-Judge Similarity Scoring

arXiv:2604.18835v1 Announce Type: cross Abstract: We propose a scalable, multifactorial experimental framework that systematically probes LLM sensitivity to subtle semantic changes in pairwise documen

Sherpa.ai Privacy-Preserving Multi-Party Entity Alignment without Intersection Disclosure for Noisy Identifiers

arXiv:2604.19219v1 Announce Type: cross Abstract: Federated Learning (FL) enables collaborative model training among multiple parties without centralizing raw data. There are two main paradigms in FL:

Sources: Micron is pushing the US Congress to pass the 'MATCH Act', which would put new export restrictions on equipment its Chinese rivals use to make chips (Karen Freifeld/Reuters)

Karen Freifeld / Reuters: Sources: Micron is pushing the US Congress to pass the “MATCH Act”, which would put new export restrictions on equipment its Chinese rivals use to make chips — Micron Technol

safetytechmeme

SpanVLA: Efficient Action Bridging and Learning from Negative-Recovery Samples for Vision-Language-Action Model

arXiv:2604.19710v1 Announce Type: new Abstract: Vision-Language-Action (VLA) models offer a promising autonomous driving paradigm for leveraging world knowledge and reasoning capabilities, especially

Stable-RAG: Mitigating Retrieval-Permutation-Induced Hallucinations in Retrieval-Augmented Generation

arXiv:2601.02993v4 Announce Type: replace Abstract: Retrieval-Augmented Generation (RAG) has become a key paradigm for reducing factual hallucinations in Large Language Models (LLMs), yet little is kn

STAR-Teaming: A Strategy-Response Multiplex Network Approach to Automated LLM Red Teaming

arXiv:2604.18976v1 Announce Type: new Abstract: While Large Language Models (LLMs) are widely used, they remain susceptible to jailbreak prompts that can elicit harmful or inappropriate responses. Thi

TEMPO: Scaling Test-time Training for Large Reasoning Models

arXiv:2604.19295v1 Announce Type: new Abstract: Test-time training (TTT) adapts model parameters on unlabeled test instances during inference time, which continuously extends capabilities beyond the r

Terrible

Terrible This is insane… The Virginia redistricting amendment on the ballot today is framed as a vote to 'restore fairness in the upcoming elections.' In reality, it turns a state that Kamala barely w

safetyelon-musk--x