Sakana AI launches Marlin product

Visual Verification Enables Inference-time Steering and Autonomous Policy Improvement

arXiv cs.AI (Artificial Intelligence) ・ 2026-06-16

VERITAS steers and self-improves robot policies at inference time

ReproRepo: Scaling Reproducibility Audits with GitHub Repository Issues

arXiv cs.CL (Computation and Language) ・ 2026-06-16

ReproRepo scales reproducibility audits using GitHub repo issues

EvolveNav: Proactive Preflection and Self-Evolving Memory for Zero-Shot Object Goal Navigation

arXiv cs.AI (Artificial Intelligence) ・ 2026-06-16

EvolveNav: a self-evolving framework for zero-shot object-goal navigation

Learning Red Agent Policy from Observations for Neurosymbolic Autonomous Cyber Agents

arXiv cs.LG (Machine Learning) ・ 2026-06-16

Learning red-agent policy from observations for cyber-defense RL

RubricsTree: Scalable and Evolving Open-Ended Evaluation of Personal Health Agents across Health Memory and Medical Skills

arXiv cs.CL (Computation and Language) ・ 2026-06-16

RubricsTree: scalable open-ended evaluation of personal health agents

DRFLOW: A Deep Research Benchmark for Personalized Workflow Prediction

arXiv cs.AI (Artificial Intelligence) ・ 2026-06-16

DRFLOW: a deep research benchmark for personalized workflow prediction

Kolmogorov Regression for Robust Diffusion Policies

arXiv cs.LG (Machine Learning) ・ 2026-06-16

Kolmogorov regression yields robust diffusion policies

All Smoke, No Alarm: Oracle Signals in Agent-Authored Test Code

arXiv cs.AI (Artificial Intelligence) ・ 2026-06-16

Study finds agent-authored test code often lacks real verification logic

Memory as a Wasting Asset: Pricing Flash Endurance for Embodied Agents, and the Limits of Doing So

arXiv cs.LG (Machine Learning) ・ 2026-06-16

Pricing flash endurance as a wasting asset for embodied agents

Your AI Travel Agent Would Book You a Bullfight: An Agentic Benchmark for Implicit Animal Welfare in Frontier AI Models

arXiv cs.CL (Computation and Language) ・ 2026-06-16

An agentic benchmark for implicit animal welfare in frontier AI

Knowledge Reutilization in Meta-Reinforcement Learning

arXiv cs.AI (Artificial Intelligence) ・ 2026-06-16

A meta-knowledge reutilization framework for meta-RL across agents

Embedded Machine Learning for Microcontroller-Class Edge Devices: Data, Feature, Evaluation, and Deployment Pipelines

arXiv cs.LG (Machine Learning) ・ 2026-06-16

A pipeline survey of embedded ML for microcontroller-class devices

Ternary Mamba: Grouped Quantization-Aware Training of W1.58A16 State Space Models

arXiv cs.LG (Machine Learning) ・ 2026-06-16

Ternary Mamba: grouped QAT for W1.58A16 state space models

Querying an astronomical database using large language models: the ALeRCE text-to-SQL system

arXiv cs.AI (Artificial Intelligence) ・ 2026-06-16

A text-to-SQL system for querying the ALeRCE astronomical database

S4oP: Operator-level Pruning of Structured State Space Models for Resource-Constrained Devices

arXiv cs.LG (Machine Learning) ・ 2026-06-16

S4oP prunes structured state space models at the operator level

Agentic AI-based Framework for Mitigating Premature Diagnostic Handoff and Silent Hallucination in Healthcare Applications

arXiv cs.AI (Artificial Intelligence) ・ 2026-06-16

A multi-agent framework against premature handoff and silent hallucination

NoiseTilt: Noise-Tilted Reverse Kernels for Diffusion Reward Alignment

arXiv cs.LG (Machine Learning) ・ 2026-06-16

NoiseTilt injects reward gradients via the noise term in diffusion

PseudoBench: Measuring How Agentic Auto-Research Fuels Pseudoscience

arXiv cs.CL (Computation and Language) ・ 2026-06-16

PseudoBench measures how agentic auto-research fuels pseudoscience

ConSA: Controllable Sparsity in Hybrid Attention via Learnable Allocation

arXiv cs.CL (Computation and Language) ・ 2026-06-16

ConSA: controllable sparsity in hybrid attention via learnable allocation

Compositional Skill Routing for LLM Agents: Decompose, Retrieve, and Compose

arXiv cs.CL (Computation and Language) ・ 2026-06-16

Compositional skill routing for LLM agents: decompose, retrieve, compose

ProvenanceGuard: Source-Aware Factuality Verification for MCP-Based LLM Agents

arXiv cs.CL (Computation and Language) ・ 2026-06-16

ProvenanceGuard: source-aware factuality verification for MCP agents

Recursive Scaling in Masked Diffusion Models

arXiv cs.LG (Machine Learning) ・ 2026-06-16

Recursive scaling in masked diffusion models

LLM Consumer Behavior Theory: Foundations of a Novel Research Field

arXiv cs.AI (Artificial Intelligence) ・ 2026-06-16

LLM Consumer Behavior Theory: a new field for agentic markets

Half a Link can Be Enough to Predict a Whole Link: Understanding Generalization in Knowledge Graph Foundation Models

arXiv cs.LG (Machine Learning) ・ 2026-06-16

Half a link can predict a whole link: generalization in KG foundation models

VoidPadding: Let [VOID] Handle Padding in Masked Diffusion Language Models so that [EOS] Can Focus on Semantic Termination

arXiv cs.CL (Computation and Language) ・ 2026-06-16

VoidPadding lets [VOID] handle padding so [EOS] focuses on termination

Differential Privacy of Gaussian Process Posterior Sampling

arXiv cs.LG (Machine Learning) ・ 2026-06-16

Differential privacy of Gaussian process posterior sampling

SoftMoE: Soft Differentiable Routing for Mixture-of-Experts in LLMs

arXiv cs.LG (Machine Learning) ・ 2026-06-16

SoftMoE: soft differentiable routing for mixture-of-experts in LLMs

Revisiting Structural Dependency in Autoregressive Multi-Task Table Recognition via Order-Independent Cell-Level Representations

arXiv cs.LG (Machine Learning) ・ 2026-06-16

Order-independent cell representations revisit table recognition

AnchorKV: Safety-Aware KV Cache Compression via Soft Penalty with a Refusal Anchor

arXiv cs.LG (Machine Learning) ・ 2026-06-16

AnchorKV: safety-aware KV cache compression via soft penalties

GameCraft-Bench: Can Agents Build Playable Games End-to-End in a Real Game Engine?

arXiv cs.CL (Computation and Language) ・ 2026-06-16

GameCraft-Bench: can agents build playable games end-to-end?

Environment-Grounded Automated Prompt Optimization for LLM Game Agents

arXiv cs.CL (Computation and Language) ・ 2026-06-16

Environment-grounded automated prompt optimization for LLM game agents

From Drift to Coherence: Stabilizing Beliefs in LLMs

arXiv cs.LG (Machine Learning) ・ 2026-06-16

From drift to coherence: stabilizing beliefs in LLMs

Improving low-resource ASR using bilingual fine-tuning with language identification: a cross-linguistic evaluation

arXiv cs.CL (Computation and Language) ・ 2026-06-16

Improving low-resource ASR via bilingual fine-tuning with language ID

A Framework for Evaluating Agentic Skills at Scale

arXiv cs.CL (Computation and Language) ・ 2026-06-16

A framework for evaluating agentic skills at scale

Position: Coding Benchmarks Are Misaligned with Agentic Software Engineering

arXiv cs.CL (Computation and Language) ・ 2026-06-16

Position: coding benchmarks are misaligned with agentic software engineering

Vision-language models for chest radiography do not always need the image

arXiv cs.CL (Computation and Language) ・ 2026-06-16

Vision-language models for chest radiography do not always need the image

EComAgentBench: Benchmarking Shopping Agents on Long-Horizon Tasks with Distributed Hidden Intent

arXiv cs.CL (Computation and Language) ・ 2026-06-16

EComAgentBench: shopping agents on long-horizon tasks with hidden intent

LLMs Infer Cultural Context but Fail to Apply It When Responding

arXiv cs.CL (Computation and Language) ・ 2026-06-16

LLMs infer cultural context but fail to apply it when responding

EnvRL: Learn from Environment Dynamics in Agentic Reinforcement Learning

arXiv cs.CL (Computation and Language) ・ 2026-06-16

EnvRL learns from environment dynamics in agentic RL

Beyond Domains: Reusing Web Skills via Transferable Interaction Patterns

arXiv cs.CL (Computation and Language) ・ 2026-06-16

Reusing web skills via transferable interaction patterns

OPD-Evolver: Cultivating Holistic Agent Evolver via On-Policy Distillation

arXiv cs.CL (Computation and Language) ・ 2026-06-16

OPD-Evolver cultivates self-evolving agents via on-policy distillation

Context-Aware RL for Agentic and Multimodal LLMs

arXiv cs.CL (Computation and Language) ・ 2026-06-15

ContextRL rewards picking the right context to ground answers

Exact Posterior Score Estimation for Solving Linear Inverse Problems

arXiv cs.LG (Machine Learning) ・ 2026-06-15

Exact closed-form posterior score for linear inverse problems

Benchmarking LLM Agents on Meta-Analysis Articles from Nature Portfolio

arXiv cs.CL (Computation and Language) ・ 2026-06-15

A benchmark for LLM agents on Nature Portfolio meta-analyses

DEEPRUBRIC: Evidence-Tree Rubric Supervision for Efficient Reinforcement Learning of Deep Research Agents

arXiv cs.CL (Computation and Language) ・ 2026-06-15

DeepRubric: evidence-tree rubrics to boost deep-research agent RL

HAMON: Passive Optical Sequence Mixing for Long-Horizon Forecasting

arXiv cs.LG (Machine Learning) ・ 2026-06-15

HAMON: a passive optical core for long-horizon forecasting

TokenPilot: Cache-Efficient Context Management for LLM Agents

arXiv cs.CL (Computation and Language) ・ 2026-06-15

TokenPilot cuts LLM-agent context costs ~61% while preserving prompt cache

TuneJury: An Open Metric for Improving Music Generation Preference Alignment

arXiv cs.LG (Machine Learning) ・ 2026-06-15

TuneJury: an open reward model for text-to-music preference

Bayesian Inference and Decision Audits for Public Archives of Frontier AI Evaluations

arXiv cs.AI (Artificial Intelligence) ・ 2026-06-15

Bayesian audit of public frontier-AI evaluation archives proposed

ActiveSAM: Image-Conditional Class Pruning for Fast and Accurate Open-Vocabulary Segmentation

arXiv cs.LG (Machine Learning) ・ 2026-06-15

ActiveSAM turns frozen SAM 3 into a training-free open-vocab segmenter

Agent trajectories as programs: fingerprinting and programming coding-agent behavior

arXiv cs.LG (Machine Learning) ・ 2026-06-15

Coding agents have behavioral fingerprints identifiable from trajectories

Dynestyx: A Probabilistic Programming Library for Dynamical Systems

arXiv cs.LG (Machine Learning) ・ 2026-06-15

Dynestyx: a probabilistic programming library with first-class SSMs

Decoupling Inference from State Updates in Low-Latency Feature Engines via Probabilistic Thinning

arXiv cs.LG (Machine Learning) ・ 2026-06-15

Probabilistic thinning decouples inference from state updates in streams

Probing Low Frame Rate Degradation in Neural Audio Codecs

arXiv cs.AI (Artificial Intelligence) ・ 2026-06-15

Probing why neural audio codecs degrade at low frame rates

Beyond the Smile: A Hybrid Convolutional VAE for Crypto Volatility Surfaces

arXiv cs.LG (Machine Learning) ・ 2026-06-15

A convolutional VAE for completing crypto implied-volatility surfaces

Phantoms and Disclosures: a Causal Framework for Auditing Synthetic Data

arXiv cs.LG (Machine Learning) ・ 2026-06-15

A causal auditing framework to detect synthetic-data privacy disclosures

A Causal Model of Theory of Mind in Conflict for Artificial Intelligence

arXiv cs.AI (Artificial Intelligence) ・ 2026-06-15

A structural causal model for when AI should engage theory of mind in conflict

Exploring Extrinsic and Intrinsic Properties for Effective Reasoning with Code Interpreter

arXiv cs.CL (Computation and Language) ・ 2026-06-15

Study probes extrinsic and intrinsic traits of code-interpreter reasoning

RAID: Semantic Graph Diffusion for True Cold-Start and Cross-Lingual Forecasting

arXiv cs.AI (Artificial Intelligence) ・ 2026-06-15

RAID: retrieval-augmented diffusion for cold-start, cross-lingual forecasting

MA-SBI: Misspecification-Aware Simulation-Based Inference via Side-Channel Guidance

arXiv cs.AI (Artificial Intelligence) ・ 2026-06-15

MA-SBI: misspecification-aware inference via side-channel guidance

Greed Is Learned: Visible Incentives as Reward-Hacking Triggers

arXiv cs.AI (Artificial Intelligence) ・ 2026-06-15

Greed Is Learned: RL agents get addicted to visible reward channels

LESS Is More: Mutual-Stability Sampling for Diffusion Language Models

arXiv cs.CL (Computation and Language) ・ 2026-06-15

LESS: a training-free adaptive sampler for diffusion language models

Binary Tracking for Spatial QA and Navigation with Open Vision-Language Models

arXiv cs.AI (Artificial Intelligence) ・ 2026-06-15

Binary Tracking: open vision-language models for spatial QA and navigation

Semantic Flip: Synthetic OOD Generation for Robust Refusal in Embodied Question Answering and Spatial Localization

arXiv cs.AI (Artificial Intelligence) ・ 2026-06-15

Semantic Flip: synthetic OOD generation for robust refusal in embodied agents

Follow the Latent Roadmap: Navigating Revocable Decoding for Diffusion LLMs with Anchor Tokens

arXiv cs.CL (Computation and Language) ・ 2026-06-15

Anchor-token roadmap for revocable decoding in diffusion LLMs

Tying the Loop -- Tied Expert Layers in Mixture-of-Experts Language Models

arXiv cs.CL (Computation and Language) ・ 2026-06-15

Paper: Expert Tying shares MoE expert params across layers

How Much Can We Trust LLM Search Agents? Measuring Endorsement Vulnerability to Web Content Manipulation

arXiv cs.CL (Computation and Language) ・ 2026-06-15

Paper: framework measures LLM search-agent endorsement risk

GIST-CMTF: Goal-State Inference for Causal Minimal Tool Filtering in LLM Agents

arXiv cs.AI (Artificial Intelligence) ・ 2026-06-15

GIST-CMTF adds goal-state inference to causal minimal tool filtering

LLM-based Visual Code Completion for Aerospace Geometric Design

arXiv cs.CL (Computation and Language) ・ 2026-06-15

Paper: LLM visual-programming copilot for aerospace design

LabOSBench: Benchmarking Computer Use Agents for Scientific Instrument Control

arXiv cs.AI (Artificial Intelligence) ・ 2026-06-15

LabOSBench: a simulated testbed for computer-use agents controlling instruments

OpenClaw-Skill: Collective Skill Tree Search for Agentic Large Language Models

arXiv cs.CL (Computation and Language) ・ 2026-06-15

OpenClaw-Skill: collective skill tree search for LLM agents

Skill-to-LoRA: From Using Skills to Learning Behaviors for Token-Efficient LLM Agents

arXiv cs.AI (Artificial Intelligence) ・ 2026-06-15

S2L replaces runtime SKILL.md text with skill-specific LoRA adapters

MyPCBench: A Benchmark for Personally Intelligent Computer-Use Agents

arXiv cs.CL (Computation and Language) ・ 2026-06-15

MyPCBench: benchmarking personal computer-use agents

Misinformation Propagation in Benign Multi-Agent Systems

arXiv cs.CL (Computation and Language) ・ 2026-06-15

Study on misinformation propagation in benign multi-agent systems

Progressive Knowledge-Guided Large Language Model Framework for Bearing Fault Diagnosis

arXiv cs.CL (Computation and Language) ・ 2026-06-15

Physics-guided multi-scale framework for bearing fault diagnosis

Multimodal Evaluator Preference Collapse: Cross-Modal Contagion in Self-Evolving Agents

arXiv cs.CL (Computation and Language) ・ 2026-06-15

Paper on evaluator preference collapse in self-evolving agents

FraudSMSWalker: Benchmarking Agentic Large Language Models for SMS-to-Webpage Fraud Detection

arXiv cs.CL (Computation and Language) ・ 2026-06-15

FraudSMSWalker benchmark targets URL-masked SMS-to-webpage fraud

VeriGraph: Towards Verifiable Data-Analytic Agents

arXiv cs.CL (Computation and Language) ・ 2026-06-15

VeriGraph: a traceable neuro-symbolic framework for verifiable data agents

SING: Synthetic Intention Graph for Scalable Active Tool Discovery in LLM Agents

arXiv cs.CL (Computation and Language) ・ 2026-06-15

SING: synthetic intention graph for scalable active tool discovery

Can LLM Agents Infer World Models? Evidence from Agentic Automata Learning

arXiv cs.CL (Computation and Language) ・ 2026-06-15

Can LLM agents infer world models? Evidence from automata learning

Can LLM Coding Agents Reason About Time Series?

arXiv cs.CL (Computation and Language) ・ 2026-06-15

Can LLM coding agents reason about time series? A benchmark study

DoubtProbe: Black-Box Jailbreak Defense via Structural Verification and Semantic Auditing

arXiv cs.CL (Computation and Language) ・ 2026-06-15

DoubtProbe: a dual-branch inference-time defense against LLM jailbreaks

daVinci-kernel: Co-Evolving Skill Selection, Summarization, and Utilization via RL for GPU Kernel Optimization

arXiv cs.CL (Computation and Language) ・ 2026-06-15

daVinci-kernel: an RL framework co-evolving skills for GPU kernel tuning