Documented Record

Research & Intelligence Archive

A permanent chronological record of empirical AI evaluations, research breakthroughs, frontier lab dispatches, and policy updates, normalized in Asia/Kolkata timezone.

Displaying 100 archived dispatchesTimestamps rendered in Asia/Kolkata (IST)
Jev: New Decision-Making Model Demonstrates Real-Time Gameplay Capabilities
Deep Research

Jev: New Decision-Making Model Demonstrates Real-Time Gameplay Capabilities

A new AI decision-making model named Jev, developed by a co-creator of ChatGPT, has been demonstrated playing the mobile game Subway Surfers in real-time. The demonstration highlights the model's ability to process visual input and execute rapid, sequential decisions in a dynamic environment.

Reddit Singularity Community·
Security Researchers Leverage Claude to Identify OpenAI Vulnerabilities
Deep Research

Security Researchers Leverage Claude to Identify OpenAI Vulnerabilities

Security researchers have successfully utilized Anthropic’s Claude AI to identify and exploit vulnerabilities within OpenAI’s systems. This cross-platform interaction highlights the growing role of LLMs in both offensive and defensive cybersecurity research.

Syndicated Media·
AI Roundtable Experiment Exposes Mechanisms of LLM Censorship
Deep Research

AI Roundtable Experiment Exposes Mechanisms of LLM Censorship

A new project titled 'Machines Arguing' has utilized a roundtable format to facilitate discussions between various AI models. The experiment notably prompted Chinese-trained LLMs to discuss the nature and limitations of their own censorship protocols, providing a rare look into the constraints governing these systems.

Reddit OpenAI Community·
Security Researchers Leverage Claude to Identify Vulnerabilities in OpenAI Systems
Deep Research

Security Researchers Leverage Claude to Identify Vulnerabilities in OpenAI Systems

Independent security researchers have utilized Anthropic’s Claude model to assist in identifying potential security vulnerabilities within OpenAI's infrastructure. This development highlights the growing trend of using advanced LLMs as force multipliers in cybersecurity research and penetration testing.

Reddit Singularity Community·
Community-Driven Optimization Boosts AMD RDNA3 Performance in llama.cpp
Deep Research

Community-Driven Optimization Boosts AMD RDNA3 Performance in llama.cpp

A specialized fork of llama.cpp has emerged, significantly enhancing inference speeds for dual AMD Radeon 7900 XTX setups. By optimizing for RDNA3 architecture, users report a performance jump from 28 to 82 tokens per second on Qwen 3.8B models.

Reddit LocalLLaMA Community·
Security Breach: Hackers Leverage Claude to Target OpenAI Systems
Deep Research

Security Breach: Hackers Leverage Claude to Target OpenAI Systems

A recent investigation by The Wall Street Journal reveals that hackers utilized Anthropic’s Claude AI model to facilitate unauthorized access to OpenAI’s internal systems. This incident highlights a growing trend where malicious actors exploit the capabilities of large language models to automate and refine cyberattack methodologies.

Syndicated Media·
Infinite-Parameter LLMs: A New Frontier in Continuous Live Learning
Deep Research

Infinite-Parameter LLMs: A New Frontier in Continuous Live Learning

A new research paper introduces a method for 'Infinite-Parameter LLMs' that utilizes evolving delta filters to allow models to learn from live data without modifying base weights. By applying dynamic, low-rank adapters on-the-fly, the architecture enables models to retain new information and correct errors in real-time.

Reddit Singularity Community·
GPT-6 Astra Demonstrates Advanced Strategic Planning in Factorio: Space Age
Deep Research

GPT-6 Astra Demonstrates Advanced Strategic Planning in Factorio: Space Age

Recent reports indicate that an iteration of the GPT-6 Astra model successfully navigated the complex logistics and strategic requirements of the Factorio: Space Age expansion. The model completed the game's progression within a 48-hour window, highlighting advancements in long-horizon planning and systematic problem-solving.

Reddit Singularity Community·
Ninfer Engine Achieves 600 tok/s Inference on Qwen3.6-35B
Deep Research

Ninfer Engine Achieves 600 tok/s Inference on Qwen3.6-35B

A user has demonstrated high-speed local inference using the Ninfer engine, achieving 600 tokens per second on the Qwen3.6-35B model. The performance was benchmarked on an NVIDIA RTX Pro 6000 workstation, highlighting the potential for rapid local execution of mid-sized language models.

Reddit LocalLLaMA Community·
Anthropic Releases Claude-Optimized GPU Kernels for Biomolecular Research
Deep Research

Anthropic Releases Claude-Optimized GPU Kernels for Biomolecular Research

Anthropic has open-sourced a suite of GPU optimizations generated by its Claude AI model, specifically designed to accelerate biomolecular modeling. These optimizations have demonstrated a 4x performance increase across more than 30 distinct models, marking a significant milestone in AI-assisted software engineering.

Reddit Singularity Community·
AMD Strix Halo Mini PC Ecosystem: September 2026 Hardware Benchmarks
Deep Research

AMD Strix Halo Mini PC Ecosystem: September 2026 Hardware Benchmarks

A comprehensive hardware sizing and performance chart for AMD Strix Halo-based Mini PCs has been released, detailing the transition from the 395 series to the upcoming 495 series. This data provides critical insights for local AI developers looking to optimize hardware footprints for edge-based LLM inference.

Reddit LocalLLaMA Community·
Anthropic Reports Claude Now Leads 26% of Internal R&D Work
Deep Research

Anthropic Reports Claude Now Leads 26% of Internal R&D Work

Anthropic has released new data indicating that its Claude AI models are now autonomously leading 26% of the company's internal research and development projects. This represents a significant acceleration in AI-driven self-improvement, rising from near-zero involvement just six months ago.

Reddit Singularity Community·
Google and UN Launch AI-Ready Data Commons for Global Statistics
Deep Research

Google and UN Launch AI-Ready Data Commons for Global Statistics

The United Nations has launched the UN System Data Commons, an open-source platform built on Google's Data Commons infrastructure. The platform utilizes AI to unify siloed global datasets into an interconnected knowledge graph, enabling natural-language querying and automated research workflows.

Google AI Blog·
Quantifying Humor: Fable 5 Leads New LLM Comedy Benchmark
Deep Research

Quantifying Humor: Fable 5 Leads New LLM Comedy Benchmark

Researchers have introduced a specialized 'Humor Arena' benchmark to evaluate the comedic capabilities of 20 different LLMs. Using a fine-tuned open-source judge, the study identifies Fable 5 as the top-performing model in generating humorous content based on 360 frozen prompts.

Reddit LocalLLaMA Community·
Initial M5 Ultra Benchmarks Surface: High-Performance Local Inference Potential
Deep Research

Initial M5 Ultra Benchmarks Surface: High-Performance Local Inference Potential

Early performance data for the M5 Ultra chip has appeared on the OMLX benchmark portal, showcasing impressive throughput for local LLM inference. The data highlights significant token generation speeds for the Qwen 3.8 27B model, signaling a potential shift in local hardware capabilities.

Reddit LocalLLaMA Community·
Developer Leverages Astra AI to Build Interactive Nuclear Fusion Simulator
Deep Research

Developer Leverages Astra AI to Build Interactive Nuclear Fusion Simulator

A developer has successfully utilized Google's Astra AI agent framework to construct a complex, science-based interactive nuclear fusion reactor laboratory in just four hours. The project, driven by a 60-page prompt, provides a 3D environment for users to manipulate reactor parameters and observe plasma physics in real-time.

Reddit Singularity Community·
Making AI-Assisted Claims Independently Challengeable: Publication Authority and a Protocol for Falsifiable Publication Records
Deep Research

Making AI-Assisted Claims Independently Challengeable: Publication Authority and a Protocol for Falsifiable Publication Records

arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website. Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.

arXiv Research Repository·
EvolveTrade: Experience-Driven Policy Refinement for Self-Evolving LLM Trading Agents
Deep Research

EvolveTrade: Experience-Driven Policy Refinement for Self-Evolving LLM Trading Agents

arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website. Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.

arXiv Research Repository·
Physics-Constrained Digital Twins for Sensor Integrity in Urban Pedestrian Flow: Detecting Stealthy False Data Injection with Conformal Guarantees
Deep Research

Physics-Constrained Digital Twins for Sensor Integrity in Urban Pedestrian Flow: Detecting Stealthy False Data Injection with Conformal Guarantees

arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website. Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.

arXiv Research Repository·
GraphEcho: Structural Redundancy and Evidence Provenance in LLM Graph Agents
Deep Research

GraphEcho: Structural Redundancy and Evidence Provenance in LLM Graph Agents

arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website. Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.

arXiv Research Repository·
A Systematic Evaluation of the COTQ Provincial Land Cover Product: Structural Consistency, Spectral Separability, and Relative Positioning Against ESA, ESRI, and Google Products
Deep Research

A Systematic Evaluation of the COTQ Provincial Land Cover Product: Structural Consistency, Spectral Separability, and Relative Positioning Against ESA, ESRI, and Google Products

arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website. Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.

arXiv Research Repository·
FairCompressAgent: An Agentic Framework for Fairness-Aware Model Compression for FPGA Deployment
Deep Research

FairCompressAgent: An Agentic Framework for Fairness-Aware Model Compression for FPGA Deployment

arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website. Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.

arXiv Research Repository·
A Four-Stage Decomposition of Word-Problem Solving and Mechanistic Fragility in LLM Math Reasoning
Deep Research

A Four-Stage Decomposition of Word-Problem Solving and Mechanistic Fragility in LLM Math Reasoning

arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website. Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.

arXiv Research Repository·
Do Frontier Models Seek Safety Evidence Before Acting?
Deep Research

Do Frontier Models Seek Safety Evidence Before Acting?

arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website. Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.

arXiv Research Repository·
ERPBench: A State-Grounded Evaluation Paradigm for Computer-Use Agents in Enterprise Software
Deep Research

ERPBench: A State-Grounded Evaluation Paradigm for Computer-Use Agents in Enterprise Software

arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website. Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.

arXiv Research Repository·
OBC-Prune: Outcome-Based Calibration for Large Reasoning Model Pruning
Deep Research

OBC-Prune: Outcome-Based Calibration for Large Reasoning Model Pruning

arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website. Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.

arXiv Research Repository·
Collaborative Memory for Multi-Agent VLM Systems
Deep Research

Collaborative Memory for Multi-Agent VLM Systems

arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website. Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.

arXiv Research Repository·
Measuring AI Leadership: Development and Validation of a Multidimensional Measure for AI-Native Organizations
Deep Research

Measuring AI Leadership: Development and Validation of a Multidimensional Measure for AI-Native Organizations

arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website. Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.

arXiv Research Repository·
Memory Has Geometry: Non-Uniform Geometric Memory for Long-Horizon Personalized AI
Deep Research

Memory Has Geometry: Non-Uniform Geometric Memory for Long-Horizon Personalized AI

arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website. Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.

arXiv Research Repository·
When to Call an LLM: A Confidence-Gated Hybrid for Cost-Effective Emotion Recognition in Conversational AI
Deep Research

When to Call an LLM: A Confidence-Gated Hybrid for Cost-Effective Emotion Recognition in Conversational AI

arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website. Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.

arXiv Research Repository·
Contiguity, Not Importance: Budgeted Repair of Stale KV Caches After Document Edits
Deep Research

Contiguity, Not Importance: Budgeted Repair of Stale KV Caches After Document Edits

arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website. Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.

arXiv Research Repository·
TuiML: Machine Learning for AI Agents
Deep Research

TuiML: Machine Learning for AI Agents

arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website. Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.

arXiv Research Repository·
RideWay: Benchmarking Efficient Task Completion for Tool-Using Language Agents
Deep Research

RideWay: Benchmarking Efficient Task Completion for Tool-Using Language Agents

arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website. Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.

arXiv Research Repository·
Multimodal Conditioning of Fine-Tuned Stable Diffusion XL for Controllable and Culturally Faithful Ulos Motif Generation
Deep Research

Multimodal Conditioning of Fine-Tuned Stable Diffusion XL for Controllable and Culturally Faithful Ulos Motif Generation

arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website. Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.

arXiv Research Repository·
The Other Half of the Memory Wall: Serving 35B MoEs from SSD with Trained Routing Prediction
Deep Research

The Other Half of the Memory Wall: Serving 35B MoEs from SSD with Trained Routing Prediction

arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website. Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.

arXiv Research Repository·
Teaching AI, Robotics, & Community: A Hubs-Based K-12 Education Framework for Reaching Rural Schools
Deep Research

Teaching AI, Robotics, & Community: A Hubs-Based K-12 Education Framework for Reaching Rural Schools

arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website. Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.

arXiv Research Repository·
AutoTuneBench: Trustworthy Measurement for Agent Auto-Tuning of LLM Serving Engines
Deep Research

AutoTuneBench: Trustworthy Measurement for Agent Auto-Tuning of LLM Serving Engines

arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website. Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.

arXiv Research Repository·
Designing Agentic AI Workflow Portfolios under Imperfect Selection and Compute Cost
Deep Research

Designing Agentic AI Workflow Portfolios under Imperfect Selection and Compute Cost

arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website. Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.

arXiv Research Repository·
Symbolic Temporal Supervision of LLM Agents Using Contracts
Deep Research

Symbolic Temporal Supervision of LLM Agents Using Contracts

arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website. Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.

arXiv Research Repository·
WFM: Wiki Foundation Model for Complex Agentic Reasoning
Deep Research

WFM: Wiki Foundation Model for Complex Agentic Reasoning

arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website. Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.

arXiv Research Repository·
REPAIR: Resolving Long-Tail Confusion in Scientific Retrievers via Fact-Verified Iterative Refinement
Deep Research

REPAIR: Resolving Long-Tail Confusion in Scientific Retrievers via Fact-Verified Iterative Refinement

arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website. Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.

arXiv Research Repository·
BENCHCOMPASS: From Scores to Signals for Training and Harness Decisions in Payment-Domain LLMs
Deep Research

BENCHCOMPASS: From Scores to Signals for Training and Harness Decisions in Payment-Domain LLMs

arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website. Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.

arXiv Research Repository·
Who Audits Whom, on What Substrate, with What Evidence? An Independence-Graded Audit Protocol for Agentic AI
Deep Research

Who Audits Whom, on What Substrate, with What Evidence? An Independence-Graded Audit Protocol for Agentic AI

arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website. Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.

arXiv Research Repository·
Building Trust in Artificial Intelligence: A Necessity for Railway Applications
Deep Research

Building Trust in Artificial Intelligence: A Necessity for Railway Applications

arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website. Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.

arXiv Research Repository·
Where Should Agents Live? Energy-Memory Characterization of Agentic AI for the Edge-Cloud Continuum
Deep Research

Where Should Agents Live? Energy-Memory Characterization of Agentic AI for the Edge-Cloud Continuum

arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website. Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.

arXiv Research Repository·
What Counts as Strategic Reasoning? A Systematic Mapping of Chess Research on Humans, Engines, and Language Models
Deep Research

What Counts as Strategic Reasoning? A Systematic Mapping of Chess Research on Humans, Engines, and Language Models

arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website. Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.

arXiv Research Repository·
Visual Compliance via Executable Safety Rule Entailment
Deep Research

Visual Compliance via Executable Safety Rule Entailment

arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website. Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.

arXiv Research Repository·
Faithful yet Collusive: Why Chain-of-Thought Monitoring Cannot Detect Collusion in LLM Pricing Agents under Oligopolistic Competition
Deep Research

Faithful yet Collusive: Why Chain-of-Thought Monitoring Cannot Detect Collusion in LLM Pricing Agents under Oligopolistic Competition

arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website. Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.

arXiv Research Repository·
Market Signal Injection: Adversarial Context Manipulation of LLM Pricing Agents
Deep Research

Market Signal Injection: Adversarial Context Manipulation of LLM Pricing Agents

arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website. Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.

arXiv Research Repository·
Cultural Competence in Context: A Large Language Model Passes the Turing Test in Finland
Deep Research

Cultural Competence in Context: A Large Language Model Passes the Turing Test in Finland

arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website. Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.

arXiv Research Repository·
HPOQuest: A Rare-Disease Diagnostic Agent Using Active Phenotype Acquisition
Deep Research

HPOQuest: A Rare-Disease Diagnostic Agent Using Active Phenotype Acquisition

arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website. Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.

arXiv Research Repository·
WetRobo: A Reproducible Robot Kit for Coding Agents in Biological Laboratories
Deep Research

WetRobo: A Reproducible Robot Kit for Coding Agents in Biological Laboratories

arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website. Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.

arXiv Research Repository·
Risk-Aware World Modeling with Flow-Guided Occupancy Evolution for Selective Trajectory Planning in Automated Driving
Deep Research

Risk-Aware World Modeling with Flow-Guided Occupancy Evolution for Selective Trajectory Planning in Automated Driving

arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website. Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.

arXiv Research Repository·
The Mirage of Calibrated Confidence: Trajectory-Independence of Verbalized Confidence in Vision-Language Models
Deep Research

The Mirage of Calibrated Confidence: Trajectory-Independence of Verbalized Confidence in Vision-Language Models

arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website. Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.

arXiv Research Repository·
Collective Loss of Control in LLM Agent Systems: An Epidemic Account of Mutation, Contagion, and Recovery
Deep Research

Collective Loss of Control in LLM Agent Systems: An Epidemic Account of Mutation, Contagion, and Recovery

arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website. Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.

arXiv Research Repository·
First Token Matters: Understanding Safety Collapse in Large Reasoning Models
Deep Research

First Token Matters: Understanding Safety Collapse in Large Reasoning Models

arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website. Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.

arXiv Research Repository·
Beyond Routine Compliance: Cunning Data Cultivates Safety Vigilance in Large Language Models
Deep Research

Beyond Routine Compliance: Cunning Data Cultivates Safety Vigilance in Large Language Models

arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website. Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.

arXiv Research Repository·
AeroWeaver: An Embodied-Agent Harness for Weaving Aerial Skills into Distributed, Adaptive Swarm Execution
Deep Research

AeroWeaver: An Embodied-Agent Harness for Weaving Aerial Skills into Distributed, Adaptive Swarm Execution

arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website. Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.

arXiv Research Repository·
TRIPROBE: Probing Task Separability Beyond Classification for XAI
Deep Research

TRIPROBE: Probing Task Separability Beyond Classification for XAI

arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website. Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.

arXiv Research Repository·
Recursive Reasoning or Statistical Extrapolation? In-Context Learning in Multi-Agent Interdependent Decision-Making
Deep Research

Recursive Reasoning or Statistical Extrapolation? In-Context Learning in Multi-Agent Interdependent Decision-Making

arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website. Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.

arXiv Research Repository·
Reasoning through Evolution: Automatic Meta-path Discovery for LLM-based Fake News Detection
Deep Research

Reasoning through Evolution: Automatic Meta-path Discovery for LLM-based Fake News Detection

arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website. Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.

arXiv Research Repository·
The Uneven Impact of Generative AI on Student Learning: Examining the Roles of Reliance, Evaluation Literacy, and Course Policy in AI-related Courses
Deep Research

The Uneven Impact of Generative AI on Student Learning: Examining the Roles of Reliance, Evaluation Literacy, and Course Policy in AI-related Courses

arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website. Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.

arXiv Research Repository·
Beyond Truncation: Rethinking LLM Decoding as Ensemble Pruning
Deep Research

Beyond Truncation: Rethinking LLM Decoding as Ensemble Pruning

arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website. Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.

arXiv Research Repository·
Which LLM is Best for Translating Natural Language Goals to PDDL
Deep Research

Which LLM is Best for Translating Natural Language Goals to PDDL

arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website. Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.

arXiv Research Repository·
Clueing up LLMs with Tool-Augmented Deductive Reasoning
Deep Research

Clueing up LLMs with Tool-Augmented Deductive Reasoning

arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website. Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.

arXiv Research Repository·
CERA-MoA: Co-Evolving Routing Mechanisms with Continually Learning LLM Agents
Deep Research

CERA-MoA: Co-Evolving Routing Mechanisms with Continually Learning LLM Agents

arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website. Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.

arXiv Research Repository·
Compositional Policy Violations: When Step-Level Compliance Fails In Agentic AI Workflows
Deep Research

Compositional Policy Violations: When Step-Level Compliance Fails In Agentic AI Workflows

arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website. Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.

arXiv Research Repository·
Infinite-Parameter LLMs: Generating and Adapting Weights from Live Data
Deep Research

Infinite-Parameter LLMs: Generating and Adapting Weights from Live Data

arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website. Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.

arXiv Research Repository·
Function Lives Where Variance Doesn't: Task-Weighted Charts of a Language Model's Computation
Deep Research

Function Lives Where Variance Doesn't: Task-Weighted Charts of a Language Model's Computation

arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website. Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.

arXiv Research Repository·
Lost in Perception: Isolating Perceptual and Reasoning Failures in Multimodal Physics and Geometry Reasoning
Deep Research

Lost in Perception: Isolating Perceptual and Reasoning Failures in Multimodal Physics and Geometry Reasoning

arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website. Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.

arXiv Research Repository·
Compiled Agency: Frontier General-Purpose Coding Agents Build Winning Game Players from Bare Interaction - from Flappy Bird to StarCraft II and Civilization
Deep Research

Compiled Agency: Frontier General-Purpose Coding Agents Build Winning Game Players from Bare Interaction - from Flappy Bird to StarCraft II and Civilization

arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website. Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.

arXiv Research Repository·
MUSE: Benchmarking Large Vision-Language Models on Multi-Modal Understanding in Situated Education
Deep Research

MUSE: Benchmarking Large Vision-Language Models on Multi-Modal Understanding in Situated Education

arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website. Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.

arXiv Research Repository·
Flag Game: A Toy Model for Mechanistic Swarm Interpretability
Deep Research

Flag Game: A Toy Model for Mechanistic Swarm Interpretability

arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website. Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.

arXiv Research Repository·
Cognitive Extensions for Dual-Process Language Agents: Memory and Self-Reflection in Interactive Environments
Deep Research

Cognitive Extensions for Dual-Process Language Agents: Memory and Self-Reflection in Interactive Environments

arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website. Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.

arXiv Research Repository·
The Death of Schema Linking? Text-to-SQL in the Age of Well-Reasoned Language Models
Deep Research

The Death of Schema Linking? Text-to-SQL in the Age of Well-Reasoned Language Models

arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website. Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.

arXiv Research Repository·
WARD: Runtime Workload-Adaptive Vision TRansformer Framework for Dependable Edge AI
Deep Research

WARD: Runtime Workload-Adaptive Vision TRansformer Framework for Dependable Edge AI

arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website. Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.

arXiv Research Repository·
Pay Only for Disagreement: Certified No-Regression Verdicts for Model Updates with Matching Label-Complexity Bounds
Deep Research

Pay Only for Disagreement: Certified No-Regression Verdicts for Model Updates with Matching Label-Complexity Bounds

arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website. Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.

arXiv Research Repository·
Structure is not mechanism: high-gain gated-FFN rows across text and genomic foundation models
Deep Research

Structure is not mechanism: high-gain gated-FFN rows across text and genomic foundation models

arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website. Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.

arXiv Research Repository·
Lecture notes on Physics Informed Neural Networks, Neural Operators, and their applications
Deep Research

Lecture notes on Physics Informed Neural Networks, Neural Operators, and their applications

arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website. Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.

arXiv Research Repository·
Scaling Articulated Rationales for MLLM-based Recommendation
Deep Research

Scaling Articulated Rationales for MLLM-based Recommendation

arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website. Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.

arXiv Research Repository·
Rethinking Domain Specialization for Open-Ended Scientific Reasoning in Astronomy Language Models
Deep Research

Rethinking Domain Specialization for Open-Ended Scientific Reasoning in Astronomy Language Models

arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website. Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.

arXiv Research Repository·
Reflect, Revise, Reuse: Training-Free Skill Evolution for GUI Agents
Deep Research

Reflect, Revise, Reuse: Training-Free Skill Evolution for GUI Agents

arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website. Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.

arXiv Research Repository·
Accelerating Diffusion Sampling via Speculative Draft Trees
Deep Research

Accelerating Diffusion Sampling via Speculative Draft Trees

arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website. Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.

arXiv Research Repository·
Confidence Comes from Experience: Experiential Confidence Estimation from Reasoning to Agents
Deep Research

Confidence Comes from Experience: Experiential Confidence Estimation from Reasoning to Agents

arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website. Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.

arXiv Research Repository·
HINT-Plan: Human Intention-Aware Robot Task Planning in Context-Rich Environments using Vision Language Models
Deep Research

HINT-Plan: Human Intention-Aware Robot Task Planning in Context-Rich Environments using Vision Language Models

arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website. Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.

arXiv Research Repository·
When AI Generates Covariates: Causal Typing and Estimand Drift in Sequential Experiments
Deep Research

When AI Generates Covariates: Causal Typing and Estimand Drift in Sequential Experiments

arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website. Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.

arXiv Research Repository·
Information Set Emulation: Causal Certificates for AI Derived EHR Features
Deep Research

Information Set Emulation: Causal Certificates for AI Derived EHR Features

arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website. Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.

arXiv Research Repository·
AI and Human Approaches to Mathematical Problem Solving
Deep Research

AI and Human Approaches to Mathematical Problem Solving

arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website. Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.

arXiv Research Repository·
SAiFE-gym: Model-based Environments for Automated Market Making with Concentrated Liquidity
Deep Research

SAiFE-gym: Model-based Environments for Automated Market Making with Concentrated Liquidity

arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website. Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.

arXiv Research Repository·
QiT: Quantum-Inspired Transformer for Visual Recognition Task
Deep Research

QiT: Quantum-Inspired Transformer for Visual Recognition Task

arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website. Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.

arXiv Research Repository·
Reflections on Trusting Trust, Revisited: Contaminating Self-Modifying AI Coding Agents with Poisoned Benchmarks
Deep Research

Reflections on Trusting Trust, Revisited: Contaminating Self-Modifying AI Coding Agents with Poisoned Benchmarks

arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website. Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.

arXiv Research Repository·
Procedural Pretraining for Molecular Property Prediction
Deep Research

Procedural Pretraining for Molecular Property Prediction

arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website. Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.

arXiv Research Repository·
Learning Nuclear Structure with AI: Radii and Collectivity
Deep Research

Learning Nuclear Structure with AI: Radii and Collectivity

arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website. Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.

arXiv Research Repository·
Lexara-RF: Reference-Free Metrics for Evaluating Conversational Visual Analytics Agents
Deep Research

Lexara-RF: Reference-Free Metrics for Evaluating Conversational Visual Analytics Agents

arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website. Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.

arXiv Research Repository·
RoboVAD: A Large Cross-Domain Evaluation Benchmark for Anomaly Detection in Robotic Arm Manipulation Videos
Deep Research

RoboVAD: A Large Cross-Domain Evaluation Benchmark for Anomaly Detection in Robotic Arm Manipulation Videos

arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website. Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.

arXiv Research Repository·
Who Judges Matters: Measuring Family-Conditioned Preference in LLM-as-Judge Panels
Deep Research

Who Judges Matters: Measuring Family-Conditioned Preference in LLM-as-Judge Panels

arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website. Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.

arXiv Research Repository·
Does AI Assistance Leave a Temporal Fingerprint? Detecting Overreliance in AI-Assisted Writing and Programming
Deep Research

Does AI Assistance Leave a Temporal Fingerprint? Detecting Overreliance in AI-Assisted Writing and Programming

arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website. Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.

arXiv Research Repository·