Search: “persistence”

Can you even imagine the nation state level attacks METR & other eval orgs will come under? Eval orgs also need world class cybersecurity as…

Can you even imagine the nation state level attacks METR & other eval orgs will come under? Eval orgs also need world class cybersecurity as well as AI talent to do this right, they are not overlappin

agentsemad-mostaque--x

Content type

All Blog X Post Paper YouTube Reddit GitHub

Clear filters

Safety

Capable but Careless: Do Computer-Use Agents Follow Contextual Integrity?

DGX agent

arXiv:2606.23189v2 Announce Type: replace-cross Abstract: Computer-use agents (CUAs) now act on a user's behalf across personal applications such as email, calendars, and to-do lists. This cross-appli

safetyarxiv-cs-cl

14 Sept 2026

ExpertHTR: Unified Handwritten Text Recognition with Multi-Task Learning and Sparse Mixture-of-Experts

model-releasesarxiv-cs-cv

arXiv:2609.12705v1 Announce Type: new Abstract: Handwritten text recognition resources are often small and distributed across collections that differ in language, script, document structure, and annot

Hidden in Rounds: Predicting the Time Cost of 802.11 Contention in Federated Learning

arXiv:2609.12903v1 Announce Type: new Abstract: Federated learning over IEEE~802.11 shares the wireless channel among clients that send model updates. We use ns-3 to measure the frame-delivery ratio a

safetyarxiv-cs-lg

I Am No One: Style-Aware Paraphrasing for Text Anonymization

arXiv:2609.12341v1 Announce Type: new Abstract: Authorship attribution models can re-identify users from seemingly anonymized text by exploiting stable stylistic fingerprints, even after explicit iden

researcharxiv-cs-cl

In-Hospital Stroke Risk-State Classification from PPG-Derived Hemodynamic Features

arXiv:2602.09328v2 Announce Type: replace Abstract: The scarcity of temporally aligned pre-event physiological data limits the study of stroke risk states before documented clinical recognition. We fo

LLM-BabyBench: Can Language Models Plan in Worlds They Can Simulate?

arXiv:2505.12135v2 Announce Type: replace-cross Abstract: When an interactive benchmark reports a single success rate for a language-model agent, it is rarely clear what that number measures. A failur

Online Material Estimation for Conditioned Diffusion Policy in Shaping Deformable Linear Objects

arXiv:2609.12634v1 Announce Type: new Abstract: Shape control of deformable linear objects (DLOs) is challenging for imitation learning because deformation behavior varies with material properties suc

safetyarxiv-cs-ro

Physical-State-Guided Diffusion Sampling for Full-Waveform Inversion

arXiv:2609.12899v1 Announce Type: new Abstract: Full waveform inversion (FWI) estimates subsurface velocity from seismic recordings, but its ill-posedness and nonlinearity make accurate reconstruction

UkisAI Swift-Qwen3.8-27B / -58.3% thinking, x1.95 speed while keeping the accuracy of xhigh

model-releasesr-localllama

Hi everybody, we post-trained Qwen 3.8 27B to be more efficient by figuring out which tokens were linked to overthinking and penalizing them without 'attacking' the reasoning length directly then fixe

UrduFactCheck: An Agentic Fact-Checking Framework for Urdu with Evidence Boosting and Benchmarking

arXiv:2505.15063v3 Announce Type: replace Abstract: The rapid adoption of Large Language Models (LLMs) has raised important concerns about the factual reliability of their outputs, particularly in low

agentsarxiv-cs-cl

When Bias Pretends to Be Truth: How Spurious Correlations Undermine Hallucination Detection in LLMs

arXiv:2511.07318v3 Announce Type: replace Abstract: Despite substantial advances, large language models (LLMs) continue to exhibit hallucinations, generating plausible yet incorrect responses. In this

Syntheses

Wiki Lint Report — 2026-09-13

linthealth-checkautomated

Automated lint: 12 errors, 54839 warnings, 3 info

13 Sept 2026

ROCm 7.2.1 + llama.cpp on mixed Vega GPUs (gfx900 Vega64 + gfx906 MI50) — got 52-57 tok/s on Gemma 3 27B Q4 with patched rocBLAS payloads [repo + Dockerfile]

I'm documenting a reproducible setup for running current llama.cpp on a mixed AMD Vega system with both: Radeon RX Vega 64 / Vega10 / gfx900 (8GB HBM2) Radeon Instinct MI50 (Vega20) / gfx906 (16GB HBM

model-releasesr-ollama

13 Sept 2026

Finishing the Task Is Not Enough: Evaluating Agent Resilience and Considerate Participation under Accumulating Challenge

arXiv:2609.10724v1 Announce Type: new Abstract: Sustained deployment of generative AI agents requires more than isolated task success. Agents must remain useful across repeated interactions, changing

12 Sept 2026

From Parameters to Answers: How LLMs Retrieve and Use Their Internal Knowledge

arXiv:2609.11859v1 Announce Type: new Abstract: How does a language model's dependence on query-routing information and target knowledge change as it answers a question? We study this question through

12 Sept 2026

Rescaling Confidence: What Scale Design Reveals About LLM Metacognition

arXiv:2603.09309v3 Announce Type: replace Abstract: Verbalized confidence, in which LLMs report a numerical certainty score, is widely used to estimate uncertainty in black-box settings, yet the confi

researcharxiv-cs-ai

12 Sept 2026

A Fragility Spectrum for Recursive Language-Model Training

arXiv:2609.11149v1 Announce Type: new Abstract: Model-generated text is finding its way back into training corpora, and there is plenty of evidence that training on such data over and over collapses o

Augustinian BabyLM: What Ostensive Definition Can and Cannot Teach a Small Language Model

arXiv:2609.11870v1 Announce Type: new Abstract: A language model normally begins training with random word embeddings: whatever 'banana' means must be learned from training corpora. I implement St. Au

Hardware

Auto-RecSys: Harnessing Autonomous Research Agents for Industry-Scale Recommender System

arXiv:2609.10922v1 Announce Type: new Abstract: Auto-research agents have shown the potential to automate hypothesis generation, experiment execution, and iterative refinement. However, scaling this p

hardwarearxiv-cs-cl

Beyond Training: A Feasibility Taxonomy for Inference-Time AI Governance

arXiv:2609.10105v1 Announce Type: cross Abstract: Compute governance today is a governance of training: the thresholds, reporting requirements, and frontier-AI regimes now in force attach to training

Context operations to architecture modelling output from large language models and evaluation criteria for their use in systems engineering design

arXiv:2609.10132v1 Announce Type: cross Abstract: The development of generative artificial intelligence resources enables opportunities of speeding up systems and engineering design work. This contrib

safetyarxiv-cs-ai

Dream4D: Lifting Camera-Controlled I2V towards Spatiotemporally Consistent 4D Generation

arXiv:2508.07769v3 Announce Type: replace Abstract: The synthesis of spatiotemporally coherent 4D content presents fundamental challenges in computer vision, requiring simultaneous modeling of high-fi

researcharxiv-cs-cv

Hierarchical Clustering Can Jointly Satisfy Richness, Consistency, and Scale Invariance

arXiv:2609.11173v1 Announce Type: new Abstract: Despite its ubiquity, clustering lacks a universally accepted definition of what is a cluster. Kleinberg's Impossibility Theorem formalizes this difficu

Leveraging Avatar Fingerprinting: A Multi-Generator Photorealistic Talking-Head Public Database and Benchmark

model-releasesarxiv-cs-cv

arXiv:2603.26934v2 Announce Type: replace Abstract: Recent advances in photorealistic avatar generation have enabled highly realistic talking-head avatars, raising security concerns regarding identity

Logit Refiner: Improving Visual Autoregressive Models via Intra-Scale Dependency Modeling

model-releasesarxiv-cs-cv

arXiv:2609.11804v1 Announce Type: new Abstract: Visual Autoregressive Models (VAR) generate images through next-scale prediction, producing all tokens within each scale in parallel. We show that this

On the Relation between Code Quality and Machine Learning Performance: A Large-scale Empirical Study

arXiv:2609.10610v1 Announce Type: cross Abstract: Context: Computational notebooks are the standard environment for machine learning (ML) development. Within the ML community, model performance is oft

ORCH: Organizational Principles Enable Collective Intelligence in Embodied AI

arXiv:2609.11737v1 Announce Type: cross Abstract: Collective intelligence depends not only on the capabilities of individual members, but also on how those members are organized. Yet artificial multi-

agentsarxiv-cs-lg

Particle GFlowNets: Rethinking Generative Marginalization Models

arXiv:2609.11538v1 Announce Type: new Abstract: Generative Marginalization Models (MaMs) have been recently introduced as efficient neural sampling models for any-order autoregressive modelling of dis

ProMediConv: Benchmarking Proactive Conversational Agents in Legal Dispute Mediation

arXiv:2609.11101v1 Announce Type: new Abstract: Dispute mediation is essential for maintaining social harmony and resilience, yet developing skilled mediators is costly and time-consuming. Existing LL

Tutorials

Rethinking Handwritten Character Recognition

arXiv:2609.10572v1 Announce Type: new Abstract: Non-Latin handwritten character recognition (HCR) remains understudied. Dominant methods consider it as generic image classification, which uses model s

tutorialsarxiv-cs-cv

RetroThinker: Enabling Retrospective Thinking in Speech LLMs

arXiv:2609.11864v1 Announce Type: cross Abstract: Speech large language models (SpeechLLMs) offer reduced latency and retain paralinguistic nuances that are typically lost in cascaded automatic speech

The Menu Is an Execution Prior: State-Path Tool Menus for Online Agents

arXiv:2609.09395v1 Announce Type: new Abstract: Language models act through tools, yet practical agents face libraries containing thousands of interfaces. We introduce the tool menu as the short, orde

Tutorials

World in World: Explore the World with World Models

arXiv:2609.11548v1 Announce Type: new Abstract: Autoregressive video world models enable interactive, long-horizon exploration, but flexible control remains challenging. Exploring a source video from

tutorialsarxiv-cs-cv

A Dominant Diffuse Phase in the Sparse Autoencoder Phase Diagram

arXiv:2609.10299v1 Announce Type: new Abstract: Sparse autoencoders (SAEs) are increasingly used to recover interpretable features from neural-network activations, yet systematic feature co-occurrence

A Translational Note on AI Safety Evaluation

arXiv:2609.06573v1 Announce Type: new Abstract: Recent studies report that automated red-teaming finds more vulnerabilities, at lower cost, than human red-teaming on standard AI safety benchmarks, and

Applications

Aegix Pulse: A Traceable Three-Stage Architecture for Personalized Content Generation and Context-Preserving Revision

arXiv:2609.07672v1 Announce Type: new Abstract: Production content-generation systems must integrate a user's immediate task, long-term brand identity, historical evidence, and revision feedback. We p

applicationsarxiv-cs-ai

Agents Trust Tools Too Much: Measuring Reliance on Unreliable Tools

arXiv:2609.05587v1 Announce Type: new Abstract: Existing evaluations of tool-using agents primarily measure whether an agent can successfully complete diverse tasks with tools. These evaluations gener

Closing the Consistency Gap: Self-Evolving Agents That Learn to Stay on Course

arXiv:2609.08832v1 Announce Type: new Abstract: Large language model (LLM)-powered agents can be accurate on average yet unreliable in production, a discrepancy that has been observed but remains larg

Compact Visuotactile World Models for Lifting: Prediction, Reward Alignment, and Force Constraints

arXiv:2609.09597v1 Announce Type: new Abstract: Accurate contact prediction is useful for robotic manipulation only if it supports effective decisions. We investigate this connection using a compact,

safetyarxiv-cs-ro

DART: A DAG-Based Reputation and Incentive Framework via Blockchain-Enabled Governance for Trustworthy LLM Multi-Agent Collaboration

arXiv:2609.05529v1 Announce Type: cross Abstract: Large language model (LLM)-based multi-agent systems (MAS) predominantly rely on centralized orchestration and lack formal verification mechanisms for

safetyarxiv-cs-ai

Deep and shallow biases in language models

arXiv:2609.09901v1 Announce Type: new Abstract: Large language models often repeatedly select the same answer even when many alternatives are plausible. Prior work treats this concentration as bias, b

safetyarxiv-cs-cl

Diamond Agent: Agentic Control of Federated HPC Resources as a Service

arXiv:2609.06181v1 Announce Type: cross Abstract: Efficiently aggregating and orchestrating computing power across heterogeneous clusters for HPC workflows faces four practical challenges: preserving

Diffs vs. Whole Files: An Empirical Comparison of Iterative Edit-Based and Direct Generation for Flutter/Dart Code Models

arXiv:2609.05779v1 Announce Type: cross Abstract: Large language models used for code editing can be trained and deployed in at least two output regimes: direct generation, where the model emits the e

Do Web Agents Investigate Before They Decide?

arXiv:2602.05354v3 Announce Type: replace Abstract: Autonomous web agents are increasingly deployed in moderation and policy enforcement, where correct decisions often depend on evidence that is not i

Explainable Deep Learning for Price-Trade Dynamics: From Black-Box Forecasts to Effective Parametric Models

arXiv:2609.06085v1 Announce Type: cross Abstract: Understanding the joint dynamics of prices and trades is central to market microstructure, where returns and order flow interact through nonlinear and

Federated Learning for Cross-Modality Medical Image Segmentation via Augmentation-Driven Generalization

arXiv:2602.20773v2 Announce Type: replace Abstract: Purpose: Developing generalizable medical image segmentation models is challenging because imaging data are distributed across institutions and diff

researcharxiv-cs-cv

Human-agent discovery of reconfigurable in-plane ferroelectric superdomain control

arXiv:2609.06887v1 Announce Type: cross Abstract: Automated experimentation is most effective when the observables, available actions, and objective are defined before the experiment starts, as is the

safetyarxiv-cs-ai