Search: “models”

LLM Wardens: Mitigating Adversarial Persuasion with Third-Party Conversational Oversight

arXiv:2605.08321v1 Announce Type: cross Abstract: LLMs are increasingly capable of persuasion, which raises the question of how to protect users against manipulation. In a preregistered user study (N=

Content type

All Blog X Post Paper YouTube Reddit GitHub

Clear filters

Model Releases

Multi-Tier Labeling and Physics-Informed Learning for Orbital Anomaly Detection at Scale

DGX agent

arXiv:2605.09790v1 Announce Type: cross Abstract: Detecting orbital anomalies, such as maneuvers, atmospheric decay, and attitude upsets, across the rapidly growing population of low-Earth-orbit (LEO)

model-releasesarxiv-cs-ai

12 May 2026

Local Ai

On Distinguishing Capability Elicitation from Capability Creation in Post-Training: A Free-Energy Perspective

arXiv:2605.08368v1 Announce Type: new Abstract: Debates about large language model post-training often treat supervised fine-tuning (SFT) as imitation and reinforcement learning (RL) as discovery. But

local-aiarxiv-cs-ai

OTora: A Unified Red Teaming Framework for Reasoning-Level Denial-of-Service in LLM Agents

arXiv:2605.08876v1 Announce Type: new Abstract: Large Language Models (LLMs) are increasingly deployed as autonomous agents that execute tool-augmented, multi-step tasks, where latency is a critical f

Parameter-Efficient Neuroevolution for Diverse LLM Generation: Quality-Diversity Optimization via Prompt Embedding Evolution

arXiv:2605.09781v1 Announce Type: cross Abstract: Large Language Models exhibit mode collapse, producing homogeneous outputs that fail to explore valid solution spaces. We present QD-LLM, a framework

PLACO: A Multi-Stage Framework for Cost-Effective Performance in Human-AI Teams

arXiv:2605.08388v1 Announce Type: new Abstract: Human-AI teams play a pivotal role in improving overall system performance when neither the human nor the model can achieve such performance on their ow

researcharxiv-cs-ai

PlantMarkerBench: A Multi-Species Benchmark for Evidence-Grounded Plant Marker Reasoning

model-releasesarxiv-cs-cl

arXiv:2605.10032v1 Announce Type: new Abstract: Cell-type-specific marker genes are fundamental to plant biology, yet existing resources primarily rely on curated databases or high-throughput studies

PrAg-PO: Prompt Augmented Policy Optimization for Robust and Diverse Mathematical Reasoning

arXiv:2602.03190v3 Announce Type: replace-cross Abstract: Reinforcement learning algorithms such as group-relative policy optimization (GRPO) have shown strong potential for improving the mathematical

Preserving Foundational Capabilities in Flow-Matching VLAs through Conservative SFT

model-releasesarxiv-cs-ro

arXiv:2605.08879v1 Announce Type: new Abstract: Unconstrained fine-tuning of flow-matching Vision-Language-Action (VLA) models drives dense parameter overwrites, degrading pre-trained capabilities. We

Process Matters more than Output for Distinguishing Humans from Machines

arXiv:2605.06524v2 Announce Type: replace Abstract: Reliable human-machine discrimination is becoming increasingly important as large language models and autonomous agents are deployed in online setti

Prompt-Activation Duality: Improving Activation Steering via Attention-Level Interventions

arXiv:2605.10664v1 Announce Type: cross Abstract: Activation steering controls language model behavior by adding directions to internal representations at inference time, but standard residual-stream

Provable Sparse Inversion and Token Relabel Enhanced One-shot Federated Learning with ViTs

arXiv:2605.10748v1 Announce Type: cross Abstract: One-Shot Federated Learning, where a central server learns a global model in a single communication round, has emerged as a promising paradigm. Howeve

researcharxiv-cs-ai

Quantum Transfer Learning Shows Improved Robustness in Low-Data Regimes

arXiv:2605.09118v1 Announce Type: cross Abstract: Transfer learning under limited data is a challenging setting, where models must adapt to new tasks with minimal supervision. Prior work has primarily

researcharxiv-cs-lg

Real vs. Semi-Simulated: Rethinking Evaluation for Treatment Effect Estimation

arXiv:2605.10430v1 Announce Type: cross Abstract: Estimating heterogeneous treatment effects with machine learning has attracted substantial attention in both academic research and industrial practice

Reasoning Compression with Mixed-Policy Distillation

arXiv:2605.08776v1 Announce Type: new Abstract: Reasoning-centric large language models (LLMs) achieve strong performance by generating intermediate reasoning trajectories, but often incur excessive t

safetyarxiv-cs-ai

Robust Server Defense Against Unreliable Clients in One-Shot Fair Collaborative Machine Learning

arXiv:2605.08616v1 Announce Type: new Abstract: Collaborative machine learning (CML) enables multiple clients to train a global model jointly in a data-distributed setting. To address data privacy and

ROM: Real-time Overthinking Mitigation via Streaming Detection and Intervention

arXiv:2603.22016v2 Announce Type: replace-cross Abstract: Large Reasoning Models (LRMs) often reach a correct solution before their long Chain-of-Thought trace ends, yet continue with redundant verifi

Route Before Retrieve: Activating Latent Routing Abilities of LLMs for RAG vs. Long-Context Selection

model-releasesarxiv-cs-cl

arXiv:2605.10235v1 Announce Type: new Abstract: Recent advances in large language models (LLMs) have expanded the context window to beyond 128K tokens, enabling long-document understanding and multi-s

S2FT: Parameter-Efficient Fine-Tuning in Sparse Spectrum Domain

arXiv:2605.08589v1 Announce Type: new Abstract: Parameter Efficient Fine-Tuning (PEFT) is a key technique for adapting a large pretrained model to downstream tasks by fine-tuning only a small number o

SayNext-Bench: Why Do LLMs Struggle with Next-Utterance Anticipation?

arXiv:2602.00327v2 Announce Type: replace Abstract: We explore the use of large language models (LLMs) for next-utterance anticipation in human dialogue. Despite recent advances in LLMs demonstrating

SciVQR: A Multidisciplinary Multimodal Benchmark for Advanced Scientific Reasoning Evaluation

arXiv:2605.10187v1 Announce Type: new Abstract: Scientific reasoning is a key aspect of human intelligence, requiring the integration of multimodal inputs, domain expertise, and multi-step inference a

SeePhys Pro: Diagnosing Modality Transfer and Blind-Training Effects in Multimodal RLVR for Physics Reasoning

arXiv:2605.09266v1 Announce Type: new Abstract: We introduce SeePhys Pro, a fine-grained modality transfer benchmark that studies whether models preserve the same reasoning capability when critical in

SleepWalk: A Three-Tier Benchmark for Stress-Testing Instruction-Guided Vision-Language Navigation

arXiv:2605.10376v1 Announce Type: new Abstract: Vision-Language Models (VLMs) have advanced rapidly in multimodal perception and language understanding, yet it remains unclear whether they can reliabl

SlimSpec: Low-Rank Draft LM-Head for Accelerated Speculative Decoding

arXiv:2605.10453v1 Announce Type: cross Abstract: Speculative decoding speeds up autoregressive generation in Large Language Models (LLMs) through a two-step procedure, where a lightweight draft model

researcharxiv-cs-cl

Steerable but Not Decodable: Function Vectors Operate Beyond the Logit Lens

arXiv:2604.02608v2 Announce Type: replace Abstract: Activation steering presupposes that task-relevant behaviors correspond to linear directions in activation space -- directions that should both stee

SWE Atlas: Benchmarking Coding Agents Beyond Issue Resolution

arXiv:2605.08366v1 Announce Type: new Abstract: We introduce SWE Atlas, a benchmark suite for coding agents spanning three professional software engineering workflows: Codebase Q&A (124 tasks), Test W

TAD: Temporal-Aware Trajectory Self-Distillation for Fast and Accurate Diffusion LLM

arXiv:2605.09536v1 Announce Type: cross Abstract: Diffusion large language models (dLLMs) offer a promising paradigm for parallel text generation, but in practice they face an accuracy-parallelism tra

researcharxiv-cs-ai

Agents

Temporal Sampling Frequency Matters: A Capacity-Aware Study of End-to-End Driving Trajectory Prediction

arXiv:2605.10388v1 Announce Type: new Abstract: End to end (E2E) autonomous driving trajectory prediction is often trained with camera frames sampled at the highest available temporal frequency, assum

agentsarxiv-cs-cv

The Differences Between Direct Alignment Algorithms are a Blur

arXiv:2502.01237v3 Announce Type: replace Abstract: Direct Alignment Algorithms (DAAs) simplify LLM alignment by directly optimizing policies, bypassing reward modeling and RL. While DAAs differ in th

To Redact, or not to Redact? A Local LLM Approach to Deliberative Process Privilege Classification

arXiv:2605.10211v1 Announce Type: cross Abstract: Government transparency laws, like the Freedom of Information (FOIA) acts in the United States and United Kingdom, and the Woo (Open Government Act) i

Tokens-per-Parameter Coverage Is Critical for Robust LLM Scaling Law Extrapolation

arXiv:2605.08541v1 Announce Type: new Abstract: Neural scaling laws approximate a language model's loss as a power-law function of parameter count N and token count D. Following Chinchilla-style compu

Towards Cross-lingual Values Judgment: A Consensus-Pluralism Perspective

arXiv:2602.17283v2 Announce Type: replace-cross Abstract: As large language models (LLMs) are employed worldwide, existing evaluation paradigms for their multilingual capabilities primarily focus on f

Two Ways to De-Bias an LLM-as-a-Judge: A Continuous-Score Comparison of Hierarchical Bayesian Calibration and Neural-ODE Score Transport

model-releasesarxiv-cs-cl

arXiv:2605.09227v1 Announce Type: new Abstract: [Abridged] Using a Large Language Model (LLM) as an automatic rater (LLM-as-a-judge) is cheap but potentially biased: some judges run lenient, others st

UFO: A Unified Flow-Oriented Framework for Robust Continual Graph Learning

arXiv:2605.09862v1 Announce Type: cross Abstract: Graph learning research has increasingly shifted toward continual graph learning (CGL), which better reflects real-world scenarios where graphs evolve

UMEDA: Unified Multi-modal Efficient Data Fusion for Privacy-Preserving Graph Federated Learning via Spectral-Gated Attention and Diffusion-Based Operator Alignment

arXiv:2605.08288v1 Announce Type: cross Abstract: Device-free localization trains models from heterogeneous wireless and visual sensors (e.g., Wi-Fi, LiDAR) distributed across edge devices. Federated

What should post-training optimize? A test-time scaling law perspective

arXiv:2605.10716v1 Announce Type: new Abstract: Large language models are increasingly deployed with test-time strategies: sample N responses, score them with a reward model or verifier, and return th

safetyarxiv-cs-lg

When Attention Beats Fourier: Multi-Scale Transformers for PDE Solving on Irregular Domains

arXiv:2605.08318v1 Announce Type: cross Abstract: We study the problem of architecture selection for deep learning models trained to solve partial differential equations (PDEs), asking when transforme

WorldReasonBench: Human-Aligned Stress Testing of Video Generators as Future World-State Predictors

arXiv:2605.10434v1 Announce Type: new Abstract: Commercial video generation systems such as Seedance2.0 and Veo3.1 have rapidly improved, strengthening the view that video generators may be evolving i

A Theory of Online Learning with Autoregressive Chain-of-Thought Reasoning

arXiv:2605.06819v1 Announce Type: new Abstract: Autoregressive generation lies at the heart of the mechanism of large language models. It can be viewed as the repeated application of a next-token gene

researcharxiv-cs-lg

Activation Differences Reveal Backdoors: A Comparison of SAE Architectures

arXiv:2605.07324v1 Announce Type: cross Abstract: Backdoor attacks on language models pose a significant threat to AI safety, where models behave normally on most inputs but exhibit harmful behavior w

safetyarxiv-cs-ai

Agentick: A Unified Benchmark for General Sequential Decision-Making Agents

arXiv:2605.06869v1 Announce Type: new Abstract: AI agent research spans a wide spectrum: from RL agents that learn from scratch to foundation model agents that leverage pre-trained knowledge, yet no u

BEAVER: An Efficient Deterministic LLM Verifier

arXiv:2512.05439v2 Announce Type: replace Abstract: As large language models (LLMs) transition from research prototypes to production systems, practitioners often need reliable methods to verify model

safetyarxiv-cs-ai

Beyond LoRA vs. Full Fine-Tuning: Gradient-Guided Optimizer Routing for LLM Adaptation

arXiv:2605.07111v1 Announce Type: cross Abstract: Recent literature on fine-tuning Large Language Models highlights a fundamental debate. While Full Fine-Tuning (FFT) provides the representational pla

Beyond the Black Box: Interpretability of Agentic AI Tool Use

arXiv:2605.06890v1 Announce Type: new Abstract: AI agents are promising for high-stakes enterprise workflows, but dependable deployment remains limited because tool-use failures are difficult to diagn

CarCrashNet: A Large-Scale Dataset and Hierarchical Neural Solver for Data-Driven Structural Crash Simulation

arXiv:2605.07098v1 Announce Type: new Abstract: Crash simulation is a cornerstone of modern vehicle development because it reduces the need for costly physical prototypes, accelerates safety-driven de

CONSIGN: Conformal Segmentation Informed by Spatial Groupings via Decomposition

arXiv:2505.14113v3 Announce Type: replace Abstract: Most machine learning-based image segmentation models produce pixel-wise confidence scores that represent the model's predicted probability for each

researcharxiv-cs-cv

CrossCult-KIBench: A Benchmark for Cross-Cultural Knowledge Insertion in MLLMs

arXiv:2605.06115v2 Announce Type: replace Abstract: Multimodal Large Language Models (MLLMs), trained primarily on English-centric data, frequently generate culturally inappropriate or misaligned resp

CSR: Infinite-Horizon Real-Time Policies with Massive Cached State Representations

arXiv:2605.07325v1 Announce Type: cross Abstract: Deploying massive large language models (LLMs) as continuous cognitive engines for robotics is bottlenecked by the time-to-first-token (TTFT) latency