Search: “models”

Repeated-Token Counting Reveals a Dissociation Between Representations and Outputs

arXiv:2605.09239v1 Announce Type: new Abstract: Large language models fail at counting repeated tokens despite strong performance on broader reasoning benchmarks. These failures are commonly attribute

Human

Content type

All Blog X Post Paper YouTube Reddit GitHub

Clear filters

Safety

Research on Security Enhancement Methods for Adversarial Robust Large Language Model Intelligent Agents for Medical Decision-Making Tasks

DGX agent

arXiv:2605.08257v1 Announce Type: cross Abstract: Motivated by the challenge to improve the adversarial robustness, security, and trust of medical decision making intelligent agents, this study develo

safetyarxiv-cs-ai

12 May 2026

Seed Hijacking of LLM Sampling and Quantum Random Number Defense

arXiv:2605.08313v1 Announce Type: cross Abstract: Large language models (LLMs) rely on deterministic pseudorandom number generators (PRNGs) for autoregressive sampling, creating a critical supply-chai

SEMASIA: A Large-Scale Dataset of Semantically Structured Latent Representations

arXiv:2605.09485v1 Announce Type: new Abstract: Latent representations learned by neural networks often exhibit semantic structure, where concept similarity is reflected by geometric proximity in embe

Sequential Membership Inference Attacks

arXiv:2602.16596v2 Announce Type: replace Abstract: Modern AI models are not static. They go through multiple updates in their lifecycles. We propose to design Sequential Membership Inference (SeMI) a

researcharxiv-cs-lg

SoccerLens: Grounded Soccer Video Understanding Beyond Accuracy

model-releasesarxiv-cs-cv

arXiv:2605.09598v1 Announce Type: new Abstract: Vision-language models (VLMs) have recently shown strong potential in soccer video understanding. However, given the high complexity of soccer videos du

Soohak: A Mathematician-Curated Benchmark for Evaluating Research-level Math Capabilities of LLMs

arXiv:2605.09063v1 Announce Type: new Abstract: Following the recent achievement of gold-medal performance on the IMO by frontier LLMs, the community is searching for the next meaningful and challengi

SPDEBench: An Extensive Benchmark for Learning Stochastic PDEs

arXiv:2505.18511v2 Announce Type: replace Abstract: Stochastic Partial Differential Equations (SPDEs) driven by random noise play a central role in modeling physical processes with rough spatio-tempor

Statistical Scouting Finds Debate-Safe but Not Debate-Useful Cases: A Matched-Ceiling Study of Open-Weight LLM Reasoning Protocols

arXiv:2605.09618v1 Announce Type: new Abstract: When should a language model answer directly, sample and vote, or engage in multi-agent debate? Recent work shows voting often explains much of the gain

The Gordian Knot for VLMs: Diagrammatic Knot Reasoning as a Hard Benchmark

arXiv:2605.09900v1 Announce Type: new Abstract: A vision-language model can look at a knot diagram and report what it sees, yet fail to act on that structure. KnotBench pairs an 858,318-image corpus f

Unmasking On-Policy Distillation: Where It Helps, Where It Hurts, and Why

arXiv:2605.10889v1 Announce Type: cross Abstract: On-policy distillation offers dense, per-token supervision for training reasoning models; however, it remains unclear under which conditions this sign

Yeti: A compact protein structure tokenizer for reconstruction and multi-modal generation

arXiv:2605.09981v1 Announce Type: cross Abstract: Multimodal models that jointly reason over protein sequences, structures, and function annotations within a unified representation hold immense potent

researcharxiv-cs-ai

Bayesian Fine-tuning in Projected Subspaces

arXiv:2605.07706v1 Announce Type: new Abstract: Low-Rank Adaptation (LoRA) enables parameter-efficient fine-tuning of large models by decomposing weight updates into low-rank matrices, significantly r

Behavior Cue Reasoning: Monitorable Reasoning Improves Efficiency and Safety through Oversight

arXiv:2605.07021v1 Announce Type: new Abstract: Reasoning in Large Language Models (LLMs) poses a challenge for oversight as many misaligned behaviors do not surface until reasoning concludes. To addr

BoHA: Blockwise Hadamard Product Adaptation for Parameter-Efficient Fine-Tuning

arXiv:2509.21637v2 Announce Type: replace Abstract: Parameter-efficient fine-tuning (PEFT) of large language models trains a small task-specific parameter set while keeping the pretrained model frozen

Can You Break RLVER? Probing Adversarial Robustness of RL-Trained Empathetic Agents

arXiv:2605.07138v1 Announce Type: new Abstract: Reinforcement learning from verifiable emotion rewards RLVER has produced language models with strong empathetic performance, evaluated on benchmarks th

Delulu: A Verified Multi-Lingual Benchmark for Code Hallucination Detection in Fill-in-the-Middle Tasks

arXiv:2605.07024v1 Announce Type: new Abstract: Large Language Models for code generation frequently produce hallucinations in Fill-in-the-Middle (FIM) tasks -- plausible but incorrect completions suc

Safety

Dynamic one-time delivery of critical data by small and sparse UAV swarms: a model problem for MARL scaling studies

arXiv:2512.09682v2 Announce Type: replace-cross Abstract: This work studies the application of Multi-Agent Reinforcement Learning (MARL) to decentralized control of unmanned aerial vehicles to relay a

safetyarxiv-cs-ai

FastOmniTMAE: Parallel Clause Learning for Scalable and Hardware-Efficient Tsetlin Embeddings

arXiv:2605.06982v1 Announce Type: new Abstract: Embedding models in natural language processing (NLP) increasingly rely on deep architectures such as BERT, while simpler models such as Word2Vec provid

GLiGuard: Schema-Conditioned Classification for LLM Safeguard

arXiv:2605.07982v1 Announce Type: new Abstract: Ensuring safe, policy-compliant outputs from large language models requires real-time content moderation that can scale across multiple safety dimension

HiDream-Studio v.01 has been released! It is fast and powerful and open-sourced on Github | Easy Install

model-releasesr-stablediffusion

HiDream-Studio v.01 was open-sourced on May 8, 2026, releasing the HiDream-O1-Image model (8B parameters) with both undistilled and distilled variants. HiDream-O1-Image is a unified image generative f

Inference Time Causal Probing in LLMs

arXiv:2605.07631v1 Announce Type: new Abstract: Causal probing methods aim to test and control how internal representations influence the behavior of generative models. In causal probing, an intervent

IntentGrasp: A Comprehensive Benchmark for Intent Understanding

arXiv:2605.06832v1 Announce Type: cross Abstract: Accurately understanding the intent behind speech, conversation, and writing is crucial to the development of helpful Large Language Model (LLM) assis

Applications

Large Video Planner Enables Generalizable Robot Control

arXiv:2512.15840v2 Announce Type: replace-cross Abstract: General-purpose robots require decision-making models that generalize across diverse tasks and environments. Recent works build robot foundati

applicationsarxiv-cs-cv

MathlibPR: Pull Request Merge-Readiness Benchmark for Formal Mathematical Libraries

arXiv:2605.07147v1 Announce Type: cross Abstract: The ecosystem of Lean and Mathlib has become the de facto standard for large language model (LLM) assisted formal reasoning with remarkable successes

MAVEN: Multi-Agent Verification-Elaboration Network with In-Step Epistemic Auditing

arXiv:2605.07646v1 Announce Type: cross Abstract: While explicit reasoning trajectories enhance model interpretability, existing paradigms often rely on monolithic chains that lack intermediate verifi

Neural Neural Scaling Laws

arXiv:2601.19831v2 Announce Type: replace-cross Abstract: Neural scaling laws predict how language model performance improves with increased training inputs. While aggregate metrics like validation lo

Rep2Text: Decoding Full Text from a Single LLM Token Representation

arXiv:2511.06571v3 Announce Type: replace-cross Abstract: Large language models (LLMs) have achieved remarkable progress across diverse tasks, yet their internal mechanisms remain largely opaque. In t

Rethinking Experience Utilization in Self-Evolving Language Model Agents

arXiv:2605.07164v1 Announce Type: new Abstract: Self-evolving agents improve by accumulating and reusing experience from past interactions. Existing work has largely focused on how experience is const

researcharxiv-cs-cl

SR^2-LoRA: Self-Rectifying Inter-layer Relations in Low-Rank Adaptation for Class-Incremental Learning

model-releasesarxiv-cs-cv

arXiv:2605.07420v1 Announce Type: cross Abstract: Pre-trained models with parameter-efficient fine-tuning (PEFT) have demonstrated promising potential for class-incremental learning (CIL), yet catastr

Unsolvability Ceiling in Multi-LLM Routing: An Empirical Study of Evaluation Artifacts

arXiv:2605.07395v1 Announce Type: cross Abstract: Efficient routing across multiple LLMs enables cost-quality tradeoffs by directing queries to the cheapest capable model. Prior work attributes routin

Are LLMs Ready for Conflict Monitoring? Empirical Evidence from West Africa

arXiv:2605.04177v1 Announce Type: new Abstract: As LLMs enter conflict monitoring, understanding systematic distortions in their outputs is critical for humanitarian accountability. We evaluate four v

From Parameter Dynamics to Risk Scoring : Quantifying Sample-Level Safety Degradation in LLM Fine-tuning

arXiv:2605.04572v1 Announce Type: cross Abstract: Safety alignment of Large Language Models (LLMs) is extremely fragile, as fine-tuning on a small number of benign samples can erase safety behaviors l

Frontier Lag: A Bibliometric Audit of Capability Misrepresentation in Academic AI Evaluation

arXiv:2605.04135v1 Announce Type: cross Abstract: Readers of applied-domain LLM capability evaluations want to know what AI systems can currently do. That literature answers a related, but consequenti

Safety

High-Fidelity Single-Image Head Modeling with Industry-Grade Topology

arXiv:2605.04524v1 Announce Type: new Abstract: We present a single-image head mesh reconstruction framework that addresses the longstanding challenge of simultaneously preserving facial identity and

safetyarxiv-cs-cv

Laundering AI Authority with Adversarial Examples

arXiv:2605.04261v1 Announce Type: cross Abstract: Vision-language models (VLMs) are increasingly deployed as trusted authorities -- fact-checking images on social media, comparing products, and modera

llm-gemini 0.31

model-releasessimon-willison

Release: llm-gemini 0.31 gemini-3.1-flash-lite is no longer a preview. Here's my write-up of the Gemini 3.1 Flash-Lite Preview model back in March. I don't believe this new non-preview model has chang

Memory as a Markov Matrix: Sample Efficient Knowledge Expansion via Token-to-Dictionary Mapping

arXiv:2605.04308v1 Announce Type: new Abstract: Continual incorporation of new knowledge is essential for the long-term evolution of large language models (LLMs). Existing approaches typically rely on

Paraphrase-Induced Output-Mode Collapse: When LLMs Break Character Under Semantically Equivalent Inputs

arXiv:2605.04665v1 Announce Type: new Abstract: When the substantive content of a request is rewritten, do large language models still answer in the format the original task asked for? We find that th

Are LLMs More Skeptical of Entertainment News?

arXiv:2605.01727v1 Announce Type: new Abstract: Large language models (LLMs) are increasingly used for automated news credibility assessment, yet it remains unclear whether they apply even-handed stan

Atomic Fact-Checking Increases Clinician Trust in Large Language Model Recommendations for Oncology Decision Support: A Randomized Controlled Trial

arXiv:2605.03916v1 Announce Type: new Abstract: Question: Does atomic fact-checking, which decomposes AI treatment recommendations into individually verifiable claims linked to source guideline docume

researcharxiv-cs-cl

Exposing LLM Safety Gaps Through Mathematical Encoding:New Attacks and Systematic Analysis

arXiv:2605.03441v1 Announce Type: cross Abstract: Large language models (LLMs) employ safety mechanisms to prevent harmful outputs, yet these defenses primarily rely on semantic pattern matching. We s

Geometric Deviation as an Unsupervised Pre-Generation Reliability Signal: Probing LLM Representations for Answerability

arXiv:2605.03196v1 Announce Type: new Abstract: A reliable language model should be able to signal, prior to generation, when a query falls outside its knowledge. We investigate whether representation

Maximizing mutual information between prompts and responses improve LLM personalization with no additional data or human oversight

arXiv:2603.19294v2 Announce Type: replace-cross Abstract: While post-training has successfully improved large language models (LLMs) across a variety of domains, these gains heavily rely on human-labe

Parameter-Efficient Distributional RL via Normalizing Flows and a Geometry-Aware Cramer Surrogate

arXiv:2505.04310v2 Announce Type: replace-cross Abstract: Distributional Reinforcement Learning (DistRL) improves upon expectation-based methods by modeling full return distributions, but standard app

Pose Tracking with a Foundation Pose Model and an Ensemble Directional Kalman Filter

arXiv:2605.03105v1 Announce Type: new Abstract: This paper introduces the ensemble directional Kalman filter (EnDKF), an ensemble-based Kalman filtering approach for pose tracking that jointly estimat

researcharxiv-cs-lg

Reward Hacking Benchmark: Measuring Exploits in LLM Agents with Tool Use

arXiv:2605.02964v1 Announce Type: new Abstract: Reinforcement learning (RL) trained language model agents with tool access are increasingly deployed in coding assistants, research tools, and autonomou

Self-Mined Hardness for Safety Fine-Tuning

arXiv:2605.03226v1 Announce Type: new Abstract: Safety fine-tuning of language models typically requires a curated adversarial dataset. We take a different approach: score each candidate prompt's diff