Multi-Agent Computer Use
Paper • 2606.01533 • Published • 6
07. IA agentica
-
-
OpenSkill: Open-World Self-Evolution for LLM Agents
Paper • 2606.06741 • Published • 152 -
Socratic-SWE: Self-Evolving Coding Agents via Trace-Derived Agent Skills
Paper • 2606.07412 • Published • 13 -
Bayesian-Agent: Posterior-Guided Skill Evolution for LLM Agent Harnesses
Paper • 2606.08348 • Published • 16 -
SlimSearcher: Training Efficiency-Aware Web Agents via Adaptive Reward Gating
Paper • 2606.07074 • Published • 13 -
Lean4Agent: Formal Modeling and Verification for Agent Workflow and Trajectory
Paper • 2606.06523 • Published • 7 -
Role-Agent: Bootstrapping LLM Agents via Dual-Role Evolution
Paper • 2606.10917 • Published • 77 -
Retrospective Harness Optimization: Improving LLM Agents via Self-Preference over Trajectory Rollouts
Paper • 2606.05922 • Published • 68 -
Flow-DPPO: Divergence Proximal Policy Optimization for Flow Matching Models
Paper • 2606.11025 • Published • 42 -
Rethinking the Divergence Regularization in LLM RL
Paper • 2606.09821 • Published • 34 -
Online Skill Learning for Web Agents via State-Grounded Dynamic Retrieval
Paper • 2606.04391 • Published • 12 -
What Should Agents Say? Action-state Communication for Efficient Multi-Agent Systems
Paper • 2606.05304 • Published • 5 -
Struct-Searcher: Agentic Structural Thinking Advances Multimodal Deep Information Seeking
Paper • 2606.07689 • Published • 7 -
Dynamic Linear Attention
Paper • 2606.10650 • Published • 9 -
Agentic Environment Engineering for Large Language Models: A Survey of Environment Modeling, Synthesis, Evaluation, and Application
Paper • 2606.12191 • Published • 73 -
Toward Generalist Autonomous Research via Hypothesis-Tree Refinement
Paper • 2606.11926 • Published • 127 -
EvoBrowseComp: Benchmarking Search Agents on Evolving Knowledge
Paper • 2606.13120 • Published • 4 -
WebChallenger: A Reliable and Efficient Generalist Web Agent
Paper • 2606.10423 • Published • 3 -
Breaking Entropy Bounds: Accelerating RL Training via MTP with Rejection Sampling
Paper • 2606.12370 • Published • 22 -
Large Language Models Are Overconfident in Their Own Responses
Paper • 2606.03437 • Published • 3 -
HarnessBridge: Learnable Bidirectional Controller for LLM Agent Harness
Paper • 2606.12882 • Published • 15 -
The Arbiter Agent: Continually Monitoring Multi-Agent Conversations to Detect Emergent Misalignment
Paper • 2606.10747 • Published • 12 -
LLM Agents Can See Code Repositories
Paper • 2606.14061 • Published • 20 -
Skip a Layer or Loop It? Learning Program-of-Layers in LLMs
Paper • 2606.06574 • Published • 25 -
HarnessX: A Composable, Adaptive, and Evolvable Agent Harness Foundry
Paper • 2606.14249 • Published • 52 -
APPO: Agentic Procedural Policy Optimization
Paper • 2606.12384 • Published • 78 -
CODA-BENCH: Can Code Agents Handle Data-Intensive Tasks?
Paper • 2606.15300 • Published • 13 -
FastContext: Training Efficient Repository Explorer for Coding Agents
Paper • 2606.14066 • Published • 95 -
PlanBench-XL: Evaluating Long-Horizon Planning of LLM Tool-Use Agents in Large-Scale Tool Ecosystems
Paper • 2606.22388 • Published • 96 -
DataClaw0: Agentic Tailoring Multimodal Data from Raw Streams
Paper • 2606.21337 • Published • 73 -
CLI-Universe: Towards Verifiable Task Synthesis Engine for Terminal Agents
Paper • 2606.22883 • Published • 38 -
Learning from Your Own Mistakes: Constructing Learnable Micro-Reflective Trajectories for Self-Distillation
Paper • 2606.18844 • Published • 21 -
Self-Compacting Language Model Agents
Paper • 2606.23525 • Published • 19 -
Causal Discovery in the Era of Agents
Paper • 2606.23608 • Published • 8 -
FastMix: Fast Data Mixture Optimization via Gradient Descent
Paper • 2606.14971 • Published • 4 -
DanceOPD: On-Policy Generative Field Distillation
Paper • 2606.27377 • Published • 84 -
OPID: On-Policy Skill Distillation for Agentic Reinforcement Learning
Paper • 2606.26790 • Published • 59 -
The Verification Horizon: No Silver Bullet for Coding Agent Rewards
Paper • 2606.26300 • Published • 54 -
PhysiFormer: Learning to Simulate Mechanics in World Space
Paper • 2606.27364 • Published • 12 -
When Does Combining Language Models Help? A Co-Failure Ceiling on Routing, Voting, and Mixture-of-Agents Across 67 Frontier Models
Paper • 2606.27288 • Published • 5 -
Are We Ready For An Agent-Native Memory System?
Paper • 2606.24775 • Published • 140 -
Plans Don't Persist: Why Context Management Is Load Bearing for LLM Agents
Paper • 2606.22953 • Published • 1 -
Forecasting Future Behavior as a Learning Task
Paper • 2606.11445 • Published • 2 -
Qwen-AgentWorld: Language World Models for General Agents
Paper • 2606.24597 • Published • 167 -
Escaping the Self-Confirmation Trap: An Execute-Distill-Verify Paradigm for Agentic Experience Learning
Paper • 2606.24428 • Published • 53 -
OpenThoughts-Agent: Data Recipes for Agentic Models
Paper • 2606.24855 • Published • 49 -
Formalizing Latent Thoughts: Four Axioms of Thought Representation in LLMs
Paper • 2606.27378 • Published • 60 -
Towards Automating Scientific Review with Google's Paper Assistant Tool
Paper • 2606.28277 • Published • 10 -
Cluster, Route, Escalate: Cascaded Framework for Cost-Aware LLM Serving
Paper • 2606.27457 • Published • 4 -
MemSyco-Bench: Benchmarking Sycophancy in Agent Memory
Paper • 2607.01071 • Published • 26 -
Multimodal Continuous Reasoning via Asymmetric Mutual Variational Learning
Paper • 2607.00461 • Published • 25 -
CausalMix: Data Mixture as Causal Inference for Language Model Training
Paper • 2607.01104 • Published • 19 -
The State-Prediction Separation Hypothesis
Paper • 2607.01218 • Published • 9 -
AutoTrainess: Teaching Language Models to Improve Language Models Autonomously
Paper • 2606.31551 • Published • 20 -
Valdi: Value Diffusion World Models
Paper • 2607.00917 • Published • 12 -
Are Performance-Optimization Benchmarks Reliably Measuring Coding Agents?
Paper • 2607.01211 • Published • 11 -
Autonomous Scientific Discovery via Iterative Meta-Reflection
Paper • 2607.01131 • Published • 6 -
When LLMs Read Tables Carelessly: Measuring and Reducing Data Referencing Errors
Paper • 2606.32029 • Published • 7 -
Evolution Fine-Tuning: Learning to Discover Across 371 Optimization Tasks
Paper • 2606.29082 • Published • 44 -
Managing Procedural Memory in LLM Agents: Control, Adaptation, and Evaluation
Paper • 2606.23127 • Published • 25 -
Little Brains, Big Feats: Exploring Compact Language Models
Paper • 2606.30062 • Published • 15 -
Hierarchical Experimentalist Agents
Paper • 2606.29315 • Published • 4 -
SkillCoach: Self-Evolving Rubrics for Evaluating and Enhancing Agentic Skill-Use
Paper • 2607.01874 • Published • 21 -
PACE: A Proxy for Agentic Capability Evaluation
Paper • 2607.02032 • Published • 18 -
When Search Agents Should Ask: DiscoBench for Clarification-Aware Deep Search
Paper • 2606.27669 • Published • 11 -
Mental World Modeling
Paper • 2607.27201 • Published • 109 -
AgentOPSD: Recursive Self-Distillation for Agentic Reinforcement Learning
Paper • 2608.05987 • Published • 104 -
ABSeeker: Training Long-Horizon Search Agents via Answer-Backtracked Credit Assignment
Paper • 2608.05102 • Published • 68 -
Weak-to-Strong On-Policy Distillation
Paper • 2607.26246 • Published • 58 -
Towards Physics of Multimodal Pretraining: Knowledge Flow, Modality Synergy, Early Unification, and Recipes
Paper • 2608.05000 • Published • 65 -
Progressive Agent Skill Generation via Reinforcement Learning
Paper • 2608.01678 • Published • 60 -
HarnessOpt-Bench: Evaluating LLMs at Harness Optimization
Paper • 2608.06301 • Published • 36 -
WorldExam: Benchmarking World Models from Apparent Appearance to Inherent Reactivity
Paper • 2608.02603 • Published • 35 -
OneDayAgent: Towards a Long-Horizon Harness for Autonomous Agents
Paper • 2608.05013 • Published • 39 -
GDPevo: Evaluating Agent Self-Evolution on Real Business Tasks
Paper • 2608.03764 • Published • 26 -
Continual Learning in Transition
Paper • 2608.06216 • Published • 26 -
AgentStream: How Well Do Self-Evolving LLM Agents Perform Under Streaming Tasks?
Paper • 2608.00155 • Published • 25 -
CADENA: Stepwise CAD Reverse Engineering
Paper • 2608.00799 • Published • 42 -
Kimi K3: Open Frontier Intelligence
Paper • 2607.24653 • Published • 536 -
Program-as-Weights: A Programming Paradigm for Fuzzy Functions
Paper • 2607.02512 • Published • 308 -
Metis: Memory Foundation Model
Paper • 2607.26760 • Published • 277 -
Harness Handbook: Making Evolving Agent Harnesses Readable,Navigable, and Editable
Paper • 2607.13285 • Published • 221 -
AREX: Towards a Recursively Self-Improving Agent for Deep Research
Paper • 2607.21461 • Published • 153 -
DataFlow-Harness: A Grounded Code-Agent Platform for Constructing Editable LLM Data Pipelines
Paper • 2607.16617 • Published • 142 -
Weak-to-Strong Generalization via Direct On-Policy Distillation
Paper • 2607.05394 • Published • 143 -
EvolvingWorld: An Open-Schema Framework for Co-Evolving Role-Play Agents and World Model in Interactive Literary World
Paper • 2607.17250 • Published • 92 -
Search Beyond What Can Be Taught: Evolving the Knowledge Boundary in Agentic Visual Generation
Paper • 2607.05382 • Published • 82 -
ResearchStudio-Reel: Automate the Last Mile of Research from Paper to Poster, Video, and Blog
Paper • 2607.04438 • Published • 61 -
Skaling: Chinchilla's Exponents Meet Kaplan's Coupling
Paper • 2608.07222 • Published • 15 -
Round-Trip Consistency: Bidirectional Diffusion Models Can Predict Their Own Rollout Errors
Paper • 2608.00675 • Published • 10 -
The Optimizer Is the Agent: Reasoning-Driven Search across Prompts, Programs, and ML Workflows
Paper • 2608.06714 • Published • 9 -
Can MLLMs Decode the Creative Leap? Introducing C4 for Cross-Concept Understanding
Paper • 2608.06501 • Published • 3 -
CEAA: A Cognitive Embodied Agents Architecture for Interactive Computing Systems
Paper • 2608.09848 • Published • 5 -
Don't Scroll Back: Missing-Evidence Memory for Streaming Dialogue Summarization
Paper • 2608.09043 • Published • 7 -
BDH-CQ: In-Context Learning with Recurrent Latent Reasoning
Paper • 2608.09888 • Published • 805 -
Agent Memory Distillation: Empowering Small LLM Agents with Hierarchical Teacher Memory
Paper • 2608.07169 • Published • 49 -
Motif 3: Technical Report
Paper • 2608.09119 • Published • 48 -
Scaling Inherently Interpretable Language Models
Paper • 2608.07594 • Published • 21 -
Stealing Reasoning Traces from Proprietary LLM APIs
Paper • 2608.09867 • Published • 121 -
Evo-Bench: Can Language Models Improve Agent Harness?
Paper • 2608.09096 • Published • 20 -
ComBodied Agents: a New Paradigm of Human-Centric Agentic AI
Paper • 2608.10915 • Published • 194 -
Co-Evolution in Agentic Systems: Toward Self-Directed Evolution Beyond Human Design
Paper • 2608.10299 • Published • 140 -
Mendel Gödel Machine: Recursive Self-Improving Coding Agents via Comparative Evolution
Paper • 2608.07645 • Published • 27 -
VibeLifeBench: Can Your Life Agent Be Proactive and Persistent in a Living World?
Paper • 2608.10875 • Published • 18 -
Decoding-Level Taboo: A Diagnostic Stress Test for LLM Robustness
Paper • 2608.09900 • Published • 11 -
SPIEval: Evaluating Large Language Models as Mobile Assistants over Scattered Personal Information
Paper • 2608.10692 • Published • 13 -
Not Worth Another Token: Marginal Value Estimation for Efficient Deep Research Agents
Paper • 2608.08389 • Published • 11 -
DistilVDR: A Compact End-to-End Visual Document Retriever via Dual-Student Distillation
Paper • 2608.10636 • Published • 11 -
DSAgentBench: Can Agents Automate End-to-End Data-Science Workflows in Real Computer Environments?
Paper • 2608.10366 • Published • 11 -
Gaze Target Estimation Anywhere with Concepts
Paper • 2608.11367 • Published • 5 -
SkillZip: Contract-Preserving Graph Compression for Scalable Agent Skill Libraries
Paper • 2608.05604 • Published • 70 -
Can LLM Agents Stick to the Script? A Benchmark for Long-Horizon Consistency in Interactive Narratives
Paper • 2608.08160 • Published • 31 -
Self-Evolving Embodied Agents via Skill-Harness Evolution
Paper • 2608.11350 • Published • 16 -
AI4AI at Test-Time: Strong-to-Weak Capability Transfer via Harnesses
Paper • 2608.12307 • Published • 120 -
DarwinX: Evolving Agent Harnesses Through Natural Selection
Paper • 2608.07545 • Published • 121 -
AutoDesign: Meta-Harness Optimization for Long-Horizon Agentic Design
Paper • 2608.13560 • Published • 68 -
Spatial Memory Agent: Experience-Grounded Procedure Memory for Spatial Intelligence
Paper • 2608.12743 • Published • 44 -
LycheeMemory V2: Efficient Long-Term Memory for LLM Agents via Semantic Segment-Level Consolidation
Paper • 2608.12990 • Published • 14 -
Knowing When to Quit: Diagnosing and Training LLMs to Abort Futile Reasoning
Paper • 2607.29211 • Published • 14 -
Marionette: Predicting World States, Rendering Geometry, Painting Appearance
Paper • 2608.14530 • Published • 34 -
Apodex Discovery: Reality Benchmarks and Environments for Evaluating and Building Discoverative Artificial Intelligence
Paper • 2608.11341 • Published • 73 -
Intern-S2-Mobius: Foundation Model with Decoupled Knowledge and Reasoning
Paper • 2608.14290 • Published • 36 -
Second Thought: Reasoning in Parallel as LLM Agents Act and Observe
Paper • 2608.13667 • Published • 18 -
Multimodal Model Diffing for Feature Discovery and Control
Paper • 2608.09928 • Published • 11 -
Agents Catching Agents: Shortcut Cascades and Benchmark Gaming in Clinical Multi-Agent Systems
Paper • 2608.03744 • Published • 6 -
A Pathway to General-Purpose Scientific AI: Multimodal Comprehension of Scientific Images
Paper • 2608.14075 • Published • 6 -
UniProbe: A Learnable Token-Level Hallucination Detector for Large VLMs using Multi-Structural Internal Representations
Paper • 2608.10835 • Published • 4 -
StateM: Reaching 95.3% Raw Accuracy, or a \$15 Frontier Run, on Terminal-Bench 2.1 via Harness Scaling
Paper • 2608.15089 • Published • 135 -
Large Discovery Models: Empirically-grounded Model-Based Open-Ended Search
Paper • 2608.15669 • Published • 64 -
Demystifying Agent Skills: Why They Work-Until They Don't
Paper • 2608.14036 • Published • 167 -
Agentic ESOpt: Fine-Tuning Long-Horizon LLM Agents with Minimal GPU Requirements
Paper • 2608.17310 • Published • 111 -
Agent Lightning v1.0: Towards Harnessed Agentic RL
Paper • 2608.17528 • Published • 35 -
SkillForge: Self-Distilling Agents for Project-Specific Issue Resolution
Paper • 2608.18933 • Published • 14 -
HarnessRisk: A Lifecycle-Oriented Benchmark for Agent Harness Safety
Paper • 2608.17597 • Published • 11 -
StartupBench: Benchmarking General-Purpose Agents on Market-Validated End-to-End Workflows
Paper • 2608.17800 • Published • 140 -
From Sequence to Structure: Relational Uncertainty Propagation for LLM Agents
Paper • 2608.16002 • Published • 7 -
The More Popular, The Harder to Forget: Adaptive Popularity for LLM Unlearning
Paper • 2608.14229 • Published • 17 -
LLMs Get Smarter from Targeted Synthetic Multilingual Data
Paper • 2608.15964 • Published • 7 -
The Embedder's Dilemma: LLMs Are Better, but at What Cost?
Paper • 2608.12875 • Published • 16 -
EnvHarness: Awakening Static Worlds for Agent Learning
Paper • 2608.19880 • Published • 219 -
TinyCast: Probabilistic Zero-Shot Forecasting with Computed Periodicity
Paper • 2608.15767 • Published • 9 -
FlowEvo: Self-Evolving Agents through the Co-Evolution of Workflows and Executable Skills
Paper • 2607.21596 • Published • 22 -
AutoSaddler: Automatic Harness Optimization with Durable Updates from Agent Execution Traces
Paper • 2608.23041 • Published • 68 -
Recursive Experiential-Working Memory Evolution for Long-Horizon Agent Harnesses
Paper • 2608.24876 • Published • 32 -
CAFE: Self-Improving Search Agents Need Co-Evolving Feedback
Paper • 2608.24794 • Published • 7 -
Automata from Agent Traces: Failure and Next-Step Prediction
Paper • 2608.23670 • Published • 6 -
Autonomous Mathematical Discovery in an Open-World Multi-Agent Environment
Paper • 2608.23691 • Published • 4 -
JIT-Agent: Scaling Harness Intelligence via Just-in-Time Harness Evolution
Paper • 2608.25593 • Published • 114 -
FrontierChallenge: Evaluating Scientific Workflow Completion
Paper • 2608.24979 • Published • 154 -
Agent-G^2: Gaussian Guidance for Agentic Reinforcement Learning
Paper • 2608.23318 • Published • 32 -
Code World Model: Coding Agent as World Brain
Paper • 2608.25927 • Published • 41 -
Video-IFBench: Evaluating Instruction Following of Multimodal LLMs in Video Understanding Scenarios
Paper • 2608.25529 • Published • 17 -
A Modular Agent for Reliable and Auditable Spatial Relation Verification in CT Scans
Paper • 2608.21140 • Published • 5 -
Real-TurnTurk: A Multimodal Turkish Corpus for Turn-Taking Prediction
Paper • 2608.22071 • Published -
PAWBench: How Far Are We from Probabilistically Aligned World Modeling?
Paper • 2608.27345 • Published • 38 -
Training Agents to Evolve with Their Harness: TaoLive Digital Avatar Agent Technical Report
Paper • 2608.15763 • Published • 54 -
CaRGo-T: Causal Reasoning Graph-of-Thought improves Multimodal Humor Comprehension
Paper • 2608.23172 • Published • 9 -
LoopArena: Benchmarking Models as Runtime Controllers for Loop Engineering
Paper • 2608.28281 • Published • 96 -
Agentic Artifact Creation: Systems, Evaluation, Principles, and Opportunities
Paper • 2608.28122 • Published • 66 -
J-Zero: Unified Challenger--Solver--Judge Co-Evolution from Zero Data
Paper • 2608.26582 • Published • 40 -
StarHarness: Evolving Harnesses with Stratified Search for Enterprise Environments
Paper • 2608.24804 • Published • 42 -
Fast Weight Attention for Continual Learning
Paper • 2608.27763 • Published • 31 -
Evaluating the Hidden Costs of Personalization in Large Language Models
Paper • 2608.28833 • Published • 30 -
WebWorld: The Browser as a World Model for Self-Improving Web Code
Paper • 2608.30530 • Published • 9 -
The Safeguard Worked. Is the LLM System Safer?
Paper • 2609.00519 • Published • 5 -
StudentSim: Training LLM-based Student Simulators
Paper • 2609.01591 • Published • 534 -
Qwen-Drive-1.0: An Initial Step towards a Vision-Language Foundation Model for Autonomous Driving
Paper • 2609.00111 • Published • 345 -
From Production Traffic to Post-Training: Building a Self-Hosted LLM That Covers the Corporate Request Mix
Paper • 2609.01572 • Published • 36 -
Evaluating Multimodal LLMs as Generalist Vision-Language-Action Agents for Drone Control: Commanding, Approaching, Tracking and Searching
Paper • 2609.01404 • Published • 32 -
Harness-of-Harness: Multi-Day Autonomous Software Development with Continual Improvement
Paper • 2609.01481 • Published • 21 -
AgentJudgeBench: A Multi-Difficulty Benchmark for Evaluating LLM Judges on Agentic Tool-Calling
Paper • 2608.26623 • Published • 20 -
E-Commerce Bench: Evaluating LLM Agents on Long-Horizon Autonomous Business Operation
Paper • 2608.30730 • Published • 17 -
Agents in the Large: Perception-Centered Architecture for Persistent Agents
Paper • 2608.30478 • Published • 12 -
Control-Data Flow Separation: Stable Prompt Optimization in Multi-Agent LLMs
Paper • 2609.00621 • Published • 8 -
Learning Where Outcomes Change:Credit-Addressable Reasoning for Multimodal Geometry
Paper • 2608.30457 • Published • 6 -
Token-Efficient Data Reasoning Agents via Adaptive Structuring of Unstructured Data
Paper • 2608.31082 • Published • 5 -
Repo-To-Skill: Distilling GitHub Repositories Into AI4AI Skills
Paper • 2609.02749 • Published • 464 -
HarnessDev: Can LLMs Create and Evolve Their Own Agent Harness?
Paper • 2609.01437 • Published • 521 -
Aspire: Can Models Self-Evolve from Vague Goals?
Paper • 2608.31111 • Published • 411 -
EarlyEval: Cheaper Agent Evaluation via Early Outcome Prediction
Paper • 2609.02783 • Published • 62 -
SolarWM: Open Data and Scalable Training for Long-Horizon Video World Models
Paper • 2609.02886 • Published • 114 -
Language Models Can Control Their Own Attention
Paper • 2609.02737 • Published • 79 -
Ignorance or Incompetence? Constructing Knowledge-Gated, Verifiable Tasks for LLM Agents
Paper • 2608.30322 • Published • 5 -
FoldingAgent: Inferring Parametric Origami Procedures from Demonstration Videos
Paper • 2609.00377 • Published • 4 -
Portfolio Risk Bounds without Cross-Asset Return Covariances: Distributional Fields from Language-Model Representations
Paper • 2608.29692 • Published • 2 -
Environment Evolution for Terminal Agents
Paper • 2609.04128 • Published • 21 -
Using Grounded Theory for Agent Behavior Analysis at Scale
Paper • 2608.30391 • Published • 19 -
Enoki: Efficient Multi-Level Hallucination Detection
Paper • 2609.00581 • Published • 28 -
Ask Before You Optimize: Dynamic Pre-Formulation Clarification for Interactive Optimization
Paper • 2609.05258 • Published • 20 -
Causal Foundation Models
Paper • 2609.03003 • Published • 28 -
Safety for Whom? Boundary-Aware Self-Distillation for Controlled LLM Safety Refusal
Paper • 2609.04482 • Published • 11 -
EmbodiedSkills: A Unified Framework for Orchestrating, Training, and Deploying VLA Agents
Paper • 2609.01281 • Published • 13 -
Omni Interaction Agent Technical Report
Paper • 2609.08977 • Published • 90 -
OpenWAM: An Open, Modular Exploration Towards Systematic World-Action Model Pretraining
Paper • 2609.07398 • Published • 81 -
Procedural Graphs: Self-Evolving Execution Structures for LLM Agents
Paper • 2609.09153 • Published • 42 -
Agentic Visual Generation: From Generative Models to Agentic Control
Paper • 2609.06758 • Published • 34 -
Steering Geometry: Validating Human Value Geometry in LLM Steering Space
Paper • 2609.06289 • Published • 30 -
Kalman Delta Networks: Uncertainty-aware Associative Memory
Paper • 2609.07816 • Published • 30 -
What LLM Trading Agents Actually Do in Production: A Six-Month, Population-Scale Record from Two Fleets
Paper • 2609.05663 • Published • 19 -
SQS: Bayesian DNN Compression through Sparse Quantized Sub-distributions
Paper • 2510.08999 • Published • 17 -
MOLE: Detecting Insider Threats in AI Agents
Paper • 2609.06966 • Published • 17 -
Graph Machine: Towards Better Pretraining via Edges
Paper • 2609.02881 • Published • 6 -
Counter-Swarm Doctrine: Containing Coordinated Agent Intrusions
Paper • 2609.06140 • Published • 3 -
Scaling Automatic Research Agents via World Models
Paper • 2608.12564 • Published • 482 -
AgentGrad: Intervention-guided Prompt Optimization for Multi Agent Systems
Paper • 2609.08572 • Published • 58 -
T1: Terminal Agent Reinforcement Learning for Long-Horizon Tasks
Paper • 2609.11042 • Published • 66 -
SWE-Bench Pro Verified: A Reliable Benchmark for Software Engineering Agents
Paper • 2609.08149 • Published • 26 -
PARSER: Read in Parallel, Reason in Depth for Long-Context LLM Agents
Paper • 2609.06702 • Published • 25 -
Φ-Bench: Can Large Language Models Engineer the Infrastructure That Powers Them?
Paper • 2609.10226 • Published • 17 -
SAEScientist-Bench: Can AI Agents Conduct Autonomous SAE Interpretability Research?
Paper • 2609.09113 • Published • 20 -
Train Smarter, Not Harder: Switching Signal-Guided Training in Active Learning
Paper • 2609.06806 • Published • 8 -
Co-Evolving Harnesses and Models: On-Policy Correction Helps Weaker Models Catch Up Where Imitation Fails
Paper • 2609.09134 • Published • 11 -
DF26: We Cannot Tell Fake From Real Anymore
Paper • 2609.07369 • Published • 5 -
SenseNova-U1.5: Towards Native Unified Visual Intelligence
Paper • 2609.11929 • Published • 173 -
SpatialBlock: Enhancing Spatial Intelligence in LVLMs via Synthetic Block-Stacking Problem
Paper • 2609.07064 • Published • 86 -
EvoSafeHarness: Evolving Model- and Domain-Specific Harnesses for Securing Agents
Paper • 2609.05903 • Published • 63 -
DRG-MAPPO: Hierarchical Dynamic Role-Graph Multi-Agent Reinforcement Learning for Cooperative Air Combat
Paper • 2609.11155 • Published • 27 -
An Open Recipe for IMO Gold: Training Nemotron for Olympiad Mathematics
Paper • 2609.10712 • Published • 42 -
Benchmark Radar: A Living Database and Search Engine for AI Benchmarks and Evaluation
Paper • 2609.11115 • Published • 220 -
StepAudio 3 Gen Technical Report
Paper • 2609.12945 • Published • 102 -
Online Learning with LLM Experts from Limited Feedback
Paper • 2609.05820 • Published • 6 -
Atria Dawn: The Dawn of Agentic Superintelligence
Paper • 2609.15818 • Published • 375 -
PhysBrain 1.5: From Vision-Language Models to Physical Foundation Models
Paper • 2609.14973 • Published • 165 -
Discovery Foundation Models: Toward Open-Ended Discovery Intelligence
Paper • 2609.15973 • Published • 33 -
Omni-Streaming Thinking
Paper • 2609.15128 • Published • 22 -
HazardAuditor: From Executable Threats to Safer Computer-Use Agents
Paper • 2609.15134 • Published • 19 -
Learning to Solve Hard Problems in RL for LLMs by Never Giving Up
Paper • 2609.13443 • Published • 15 -
When Agents Slow Down: Understanding LLM Agents' Test-Time Strategies via Elo-per-token Analysis
Paper • 2609.15309 • Published • 13 -
Enabling Creative Exploration for Vibe Design Agents
Paper • 2609.15078 • Published • 9 -
Dynin-Robotics: Omnimodal Unified Diffusion Vision-Language-Action Model
Paper • 2609.13053 • Published • 9 -
Pick Your Poison: Learning to Select Poison Sets for Stronger LLM Backdoor Attacks
Paper • 2609.15029 • Published • 8 -
Expert-Space Exploration in MoE Reinforcement Learning
Paper • 2609.13058 • Published • 9 -
Continual Learning Mechanisms Compose for Long-Horizon Memorization
Paper • 2609.06986 • Published • 370 -
Generalized Agent Iteration: One Formal Framework for Iterative Policy Improvement and Recursive Self-Improvement
Paper • 2609.13406 • Published • 109 -
The Last AI Built by Humans: Toward Genuine Recursive Self-Improvement
Paper • 2609.11873 • Published • 82 -
Another Blueprint In The Wall: How to Ask Frontier AI Like a Kid?
Paper • 2609.14803 • Published • 12 -
Convergent Emergence of In-Context Learning Across Modalities
Paper • 2609.14011 • Published • 3 -
ScienceIDE: Turning World's Scientific Codebase into Agent Learnable Environments
Paper • 2609.19134 • Published • 106 -
Rethinking Critic Learning in PPO: Understanding and Mitigating Value Flattening
Paper • 2609.18708 • Published • 84 -
ProgramDistill: From Interactive Web Apps to Verifiable Reference-Guided SWE Tasks
Paper • 2609.18805 • Published • 67 -
Agora: Git as Shared Memory for Collective AutoResearch
Paper • 2609.18094 • Published • 58 -
EvolveTrade: Experience-Driven Policy Refinement for Self-Evolving LLM Trading Agents
Paper • 2609.17632 • Published • 44 -
HypoEvolve: Genetic Algorithms Enable Multi-Agent LLMs to Discover Scientific Hypotheses
Paper • 2609.15938 • Published • 30 -
DeepSeek-V4.1-Flash: Pushing the Limits of KV Cache Compression
Paper • 2609.19969 • Published • 235 -
SoL-Pi: Recursively Scaling Auto-Research Loops for Efficient Agent Harness
Paper • 2609.20519 • Published • 142 -
JEPA-Anything: Learning Predictive Models across Different Worlds
Paper • 2609.20800 • Published • 79 -
An Empirical Study of Harness Design for Coding Agents
Paper • 2609.20804 • Published • 95 -
Self-Evolving Search Index
Paper • 2609.19656 • Published • 63 -
When2Think: Learning When and How Much to Reason
Paper • 2609.19671 • Published • 54 -
Don't Mask the Environment: Observation Supervision Changes How Agents Explore Under RL
Paper • 2609.20715 • Published • 45 -
PACT: Can Enterprise AI Assistants Be Trusted Under Pressure?
Paper • 2609.18605 • Published • 41 -
Sample Count Is Not Enough: Candidate-Generation Strategy Shapes the Energy and Performance of LLM Test-Time Scaling
Paper • 2609.19499 • Published • 36 -
EvoOntology: A Self-Evolving Ontology Layer for Data Agents
Paper • 2609.15779 • Published • 130 -
OmniVBench: A Benchmark and Large-Scale Dataset for Omni Reference-to-Video Generation
Paper • 2609.22069 • Published • 38 -
BI-Agent and BI-Bench: Towards Automating End-to-End Business Intelligence
Paper • 2609.20886 • Published • 30 -
GraphSkillEvo: Evolutionary Optimization of Graph-Structured Agent Skills
Paper • 2609.21749 • Published • 20 -
From Pretraining to Proficiency: Real-World Subtask RL for Long-Horizon Manipulation with Minimal Human Intervention
Paper • 2609.21788 • Published • 13 -
Gricea: An Open Science Platform for Conversational AI Research
Paper • 2609.22039 • Published • 13 -
RRSI: Regularized Recursive Self-Improvement of Agent Harnesses
Paper • 2609.24972 • Published • 220 -
Harness-Zero: Harness Distillation via Agent-as-Harness
Paper • 2609.24974 • Published • 38 -
Jev-Mem: System-One-Controlled Agentic Memory for Efficient AI Agents
Paper • 2609.23986 • Published • 31 -
ShieldVLA: Feasibility-Aware Safety Alignment for Vision-Language-Action Models
Paper • 2609.13231 • Published • 19 -
1% of Tokens Can Be Enough: On Gradient Estimation in On-Policy Distillation
Paper • 2609.24432 • Published • 16 -
A Lie Detector Test for Language Models: Reading Knowledge a Model Won't Reveal
Paper • 2609.21996 • Published • 9 -
Measuring the Checker: Mutation Analysis for GPU-Kernel Benchmark Oracles
Paper • 2609.22220 • Published • 7 -
The information geometry of large language models is shared, learned, and controllable
Paper • 2609.11063 • Published • 6 -
The Tasteful Agent: Measuring and Improving Taste in Long-Horizon Tasks
Paper • 2609.25804 • Published • 165 -
Bellman Policy Optimization
Paper • 2609.15987 • Published • 49 -
Circuit Hypernetworks for Quantum-Augmented Diffusion Language Models
Paper • 2609.24657 • Published • 40 -
JEV-as-a-Judge: Accept When Confident, Escalate When Unsure
Paper • 2609.26550 • Published • 44 -
Agensh: Scaling Organizational Intelligence to 1,024 Agents
Paper • 2609.26781 • Published • 30 -
Recursive self-improvement of AI research agents
Paper • 2609.26457 • Published • 28 -
HappyWorld-Bench
Paper • 2609.24308 • Published • 60 -
Just-in-Time Memory: Learning to Curate Task-Adaptive Memory for LLM Agents
Paper • 2609.27334 • Published • 57 -
Capable yet Parsimonious: Extracting and Characterizing Hidden Chain-of-Thought in Frontier Models
Paper • 2609.26637 • Published • 27 -
PackLab: A Comprehensive Framework for Developing, Training, and Evaluating MLLMs in Robotic Bin Packing
Paper • 2609.23784 • Published • 16 -
WhatWorkedBench: Benchmarking Experimental Understanding in AI Agents
Paper • 2609.27490 • Published • 15 -
Uranus: Building the Next-Generation Simulation Infrastructure for Embodied AI
Paper • 2609.24815 • Published • 14 -
Calibration as a First-Class Criterion in LLM Evaluation
Paper • 2609.26489 • Published • 4 -
Agent-Editing World Model: Rethinking World Modeling for LLM Agents
Paper • 2609.28416 • Published • 45 -
Qwen-Planner-Agent: A Closed-Loop AI-for-AI Framework for Real-World Mobile Planner Agents
Paper • 2609.29892 • Published • 34 -
IterSynth: Rethinking Deep Search Agents via Role-Decoupled Iterative Synthesis
Paper • 2609.29444 • Published • 21 -
AgentKernel: The Trust-Native Agentic Operating System
Paper • 2609.29647 • Published • 16 -
World Action Agent: Harnessing VLMs for Robot Manipulation via World Action Rehearsal
Paper • 2609.29964 • Published • 15 -
Disaggregated Quantization: Specializing LLM Prefill and Decode
Paper • 2609.26333 • Published • 95 -
TimeEvo: Failure-Driven Self-Evolution of a Time Series Agent
Paper • 2609.27277 • Published • 33 -
LastOPD: Taming Collapse in Latent On-Policy Distillation
Paper • 2609.28845 • Published • 31 -
D-JEPA: A Decision-Aligned Latent World Model
Paper • 2609.24749 • Published • 18 -
Game Arena: Strategic LLM Evaluation in Competitive Environments
Paper • 2609.31473 • Published • 12 -
TraceDance: An Automated System for Building Agent Behavior Benchmarks from Real-World Agent Deployment Traces
Paper • 2609.33295 • Published • 72 -
How Far Are We from Removing the Visual Encoder? Scaling Laws for Encoder-Free Multimodal Pretraining
Paper • 2609.35457 • Published • 68 -
QwenGyre: An Elastic Reinforcement Learning Framework for Training xLong-Horizon Agents
Paper • 2609.33848 • Published • 51 -
Knowing When Thinking Is Not Enough: Teaching Small Reasoning Models to Reason Beyond Their Parametric Knowledge
Paper • 2609.34327 • Published • 41 -
Learning to Learn from Context: Synthetic Training from Perturbed Public Documents
Paper • 2609.33642 • Published • 35 -
Nereus: Adaptive Parallelism for LLM Post-Training
Paper • 2609.34645 • Published • 20 -
BaRe-Mem: Bayesian Reliability Memory for Robust and Adaptive Agent Consultation
Paper • 2609.35551 • Published • 8 -
Relic: From Multi-Agent Collaboration to Persistent Organizational Capability
Paper • 2609.32965 • Published • 7 -
Reinforcing Agentic Creativity in Scientific Ideation with Night Science
Paper • 2609.35706 • Published • 5 -
Not All Objectives Are Born Equal: Priority-Constrained Descent for Hierarchical Multi-Objective Optimization
Paper • 2606.29521 • Published • 5 -
When Do Model Internals Help? Exploring the Role of Representation Engineering in LLM Safety
Paper • 2609.34771 • Published • 4 -
Raven: The Harness of Harnesses for Composable Agentic Intelligence
Paper • 2609.33439 • Published • 711 -
Omni-IO Skills: Harnessing Your Agent Omni-Native
Paper • 2609.31847 • Published • 485 -
MaLiang-Harness: A Programmable Path to Image and Video Generation
Paper • 2609.34309 • Published • 219 -
What Makes World Action Models Generalize? An Empirical Study of Test-Time Future Modeling
Paper • 2609.34981 • Published • 141 -
Think Before You Score: Thinking Reward Model for Visual Generation
Paper • 2609.37372 • Published • 104 -
LLMs are General Asynchronous Agents
Paper • 2609.35427 • Published • 86 -
Follow the Entities: A Corpus Map for Agentic Search
Paper • 2609.37226 • Published • 101 -
Context Language Models
Paper • 2609.37725 • Published • 47 -
Marathoner: Ultra-Long-Horizon Autonomous Intelligence
Paper • 2609.34378 • Published • 43 -
Omni-Decision: Evidence-Ledger Planning for Omni-Modal Agents
Paper • 2607.11433 • Published • 31 -
When Does Dense Retrieval Need Asymmetric Geometry? A Bias-Variance Theory of Shared and Dual Projections
Paper • 2609.32488 • Published • 27 -
Chinese-Jev: Bringing System One Model to Chinese-Language Tasks
Paper • 2609.36965 • Published • 26 -
Routing Should Pay for Itself: Sparse Supervision for Economical LLM Routing
Paper • 2609.37402 • Published • 19 -
Can Agents Design Libraries for Agents?
Paper • 2609.36730 • Published • 12 -
Where the Model Changes Its Mind: Hindsight-Divergence Localization for Efficient Reinforcement Learning with Verifiable Rewards
Paper • 2609.36864 • Published • 10 -
False Frontiers: Diagnosing and Mitigating Co-Cheating in Self-Evolving Search Agents
Paper • 2609.39102 • Published • 234 -
AREX-2: Advancing Self-Improving Agents through Long-Horizon Reflective Tasks
Paper • 2609.38288 • Published • 117 -
EvoDuet: Bilevel Co-Evolution of Web Searching and Task Solving for Scientific Discovery
Paper • 2609.40340 • Published • 112 -
EVOKE: Eliciting World Knowledge in Agents for Transferable Decision-Making
Paper • 2609.38334 • Published • 75 -
Thinking Outside the Box: Can Language Models Rely on External Guidance Selectively?
Paper • 2609.39578 • Published • 72 -
OSWorld-Science: A Benchmark of Computer Use Agents for Learning and Using Scientific Software
Paper • 2609.39903 • Published • 41 -
AIM: Agentic Idea Management for Automated Research
Paper • 2609.38445 • Published • 53 -
LANTERN: Illuminating Hidden Mathematical Knowledge in Language Models
Paper • 2609.32264 • Published • 51 -
Scaling Laws for Looped Mixture of Experts
Paper • 2609.40316 • Published • 24 -
PivotOPD: Learning to Recover from Pivotal Mistakes in Multi-Turn Agents
Paper • 2609.40285 • Published • 25 -
MILO: Automated Harness Discovery via Orchestrated Multi-Agent Evolution
Paper • 2609.38349 • Published • 24 -
The Evolution of Attention in Large Language Models: Mechanisms, Trade-offs, and Emerging Trends
Paper • 2609.39661 • Published • 14 -
AdviSD: Learning to Advise Frontier LLMs via Targeted Multi-Turn Self-Distillation
Paper • 2609.38142 • Published • 14 -
SeLMRoute: Probabilistic Semantic Evidence for Large Language Model Routing
Paper • 2609.34736 • Published • 13 -
SkillGym: Training Skill-Use Agents with Automatic Verifiable Environment Generation
Paper • 2609.37539 • Published • 14 -
PatchHolmes: Agentic Patch Retrieval via Listwise Selection
Paper • 2609.38807 • Published • 13 -
Safety of Latent Communication in Multi-Agent Systems
Paper • 2609.39788 • Published • 10 -
Overcoming Scaling Limits in On-Policy Self-Distillation for LLM Reasoning
Paper • 2609.37915 • Published • 10 -
NavHarness: Towards Lifelong Embodied Navigation
Paper • 2609.34276 • Published • 6 -
Training LLM Judges from Language Feedback via Position-Selective Self-Distillation
Paper • 2609.38792 • Published • 6 -
SkillSeek: Revisiting Agent Skill Retrieval at Marketplace Scale
Paper • 2609.38822 • Published • 7 -
Not Every Token Is Worth Distilling: Selective Supervision for Direct-OPD
Paper • 2609.29142 • Published • 6 -
Synthetic Pre-pretraining Survives Scale, but Not as a Grammatical Prior
Paper • 2609.39827 • Published • 13 -
How Much Is an AI Token Worth? Scaling Laws for Wild AI-Generated Web Text
Paper • 2609.40295 • Published • 5 -
Evaluating Bounded Autonomy in Regulated Agentic AI: A Diagnostic Harness with Constitutional Rewards, Escalation Labels, and Runtime Governance
Paper • 2609.37501 • Published • 4 -
ATLAS: Aligned Transport of Latent Structure for Reliable World Model Planning
Paper • 2609.36333 • Published • 4 -
EviRover: Reinforcing Agentic Perception Beyond a Glance
Paper • 2609.40230 • Published • 4 -
The Geometry of Inference in Transformer Residual Streams
Paper • 2609.37824 • Published • 3 -
GraphForge: Training Working Agents with Graph-Anchored Workspace Synthesis
Paper • 2609.38923 • Published • 77 -
Adaptive Reward Routing: Dynamic Multi-Reward Optimization for Joint Audio-Video Diffusion via Forward-Process RL
Paper • 2609.37200 • Published • 140 -
Agent Priors-guided Policy Learning
Paper • 2609.35690 • Published • 88 -
ActiveSaddler: Automated Curriculum Learning for Agent Harness Optimization
Paper • 2610.00906 • Published • 87 -
Decentralized Master-Mind: Joint Action Refinement through Iterative Intent Denoising in Multi-Agent Pathfinding
Paper • 2609.32019 • Published • 66 -
AutoDataBench: A Data-centric Testbed for Accelerating Auto Research
Paper • 2609.40097 • Published • 50 -
Argo-Bench: Evaluating Data Agents on Enterprise-Scale Workflows
Paper • 2610.02122 • Published • 33 -
Smaller Models, Better Rejects: Preference Distillation Scaling
Paper • 2609.38987 • Published • 32 -
DataMagic: Authoring Data Videos through Declarative Multi-Agent Orchestration
Paper • 2609.33403 • Published • 17 -
AgSpec: Pushing the Limits of Retrieval-Based Speculative Decoding in Coding Agent Pipelines
Paper • 2610.01108 • Published • 15 -
Memorizon: Training World Models Beyond Their Context Window
Paper • 2610.00544 • Published • 12 -
KaliBench: A Fine-Grained Benchmark for Cybersecurity Tool Use on Kali Linux with Runtime-Free Verifiable Rewards
Paper • 2610.02206 • Published • 10 -
Pay for the Fault, Not the Flow: Label-Free In-Flow Multi-Agent Workflow Optimization
Paper • 2610.01017 • Published • 8 -
Scaling Trajectories for Complex Tasks through Recursive Self-Rewrite
Paper • 2610.02826 • Published • 106 -
LexReward: A Taxonomy-Driven Reward Framework for Legal Language Models
Paper • 2609.39071 • Published • 69 -
HyperBrowseComp: A Multilingual and Multimodal Stress Test for Web-Browsing Agents
Paper • 2610.03574 • Published • 60 -
Science or Slop?: Benchmarking and Mitigating Scientific Slop in AI-Generated Papers
Paper • 2610.00531 • Published • 63 -
VeriHarness: Scaling Agentic Verification for Long-Horizon Tasks
Paper • 2610.00972 • Published • 63 -
SimuVerity: Benchmarking Agents for Engineering-Grade Simulink Model Generation
Paper • 2610.02304 • Published • 56 -
Science Utopia? Closed-Loop LLM Simulation of Academic Research Ecosystems
Paper • 2610.01257 • Published • 48 -
FrugalEvo: Towards Cost-Aware LLM-Guided Program Evolution
Paper • 2610.03675 • Published • 29 -
Local Support Learning
Paper • 2610.02126 • Published • 28 -
Beyond Future Prediction: Denoising as Generative Adaptation for Robot Control
Paper • 2609.28339 • Published • 30 -
Collective Bias Mitigation via Model Routing and Collaboration
Paper • 2610.03240 • Published • 20 -
Covert Assistance: Helpful LLM Agents Evade Oversight in Multi-Agent Systems
Paper • 2609.39050 • Published • 21 -
Can Computation from Earlier Problems Help LLMs Solve New Ones?
Paper • 2609.39394 • Published • 11 -
Improving Atomic-Fact Recall via Focused Views in Unstructured Knowledge Editing
Paper • 2610.02772 • Published • 7 -
CUAWright: A Minimal Unified Interface for Digital Agents
Paper • 2610.04116 • Published • 6 -
LoGRA: Scaling LLM Reinforcement Learning with Low-Rank Gradient Sketches
Paper • 2610.06647 • Published • 59 -
ALoDLM: Adaptively Looped Diffusion Language Models
Paper • 2610.04198 • Published • 81 -
Foundations of Proactive Agents: Principles, Technical Layers, and Proactivity-Gym
Paper • 2609.37267 • Published • 40 -
LMBuild: Evaluating LLM Agents for Generating Buildable and Functional Structures
Paper • 2610.04292 • Published • 41 -
Representation-Space MMD for Diffusion Language Models
Paper • 2610.06648 • Published • 33 -
Towards Looped Models Done Right, Part II: Rethinking at Fixed Points
Paper • 2610.06833 • Published • 37 -
OSWorld-Pro: Process-based Evaluation for Computer Use Agents
Paper • 2609.24890 • Published • 28 -
Self-Generated Feedback Destabilizes Test-Time Training: A Causal Decomposition of Long-Horizon Adaptation
Paper • 2610.05076 • Published • 26 -
Optimizing the Optimizer: Language Models Discover Faster Molecular Relaxation
Paper • 2610.06577 • Published • 24 -
SearchJev: A Fast and Calibrated System-1 Model for Search Agents
Paper • 2610.05107 • Published • 40 -
Base Models Can Reason By Taking a Cue From Training Data
Paper • 2610.06851 • Published • 28 -
QuantCode Model: Specializing Language Models for Executable Algorithmic Trading Code
Paper • 2609.39420 • Published • 20 -
Capability-Driven Self-Evolution of Agent Memory
Paper • 2610.06361 • Published • 20 -
ProgressCompass: Embodied Progress Reward Models Are Lost Without the Right Context
Paper • 2609.36684 • Published • 20 -
Noise Out, Bias In: Targeted Bias Injection in Diffusion Language Models via Closed-Loop Activation Steering
Paper • 2610.05894 • Published • 17 -
LLM-as-Jev: LLMs Are Already Jev-Style Decision Models -- When and How to Fine-Tune Them
Paper • 2610.02076 • Published • 16 -
Prism: Dynamic Sparse Attention for Native 2K Joint Video-Audio Generation Model Training
Paper • 2610.05416 • Published • 23 -
What Matters for Latent Reasoning with Flow Matching
Paper • 2610.06666 • Published • 14 -
How to Loop MoE: Flatten the Experts, Untie the Attention
Paper • 2609.35751 • Published • 14 -
The Missing Primitive: Diagnosing and Repairing Mathematical Reasoning in Large Language Models
Paper • 2610.02191 • Published • 16 -
Dynamic Harness Search: Building Multi-Agent Systems Per-Query via Prediction
Paper • 2610.04137 • Published • 12 -
Periscope: Extending Frozen Language Models Beyond Their Context Window
Paper • 2610.04047 • Published • 9 -
Training Numerical Intelligence via Auto-Diagnosis and Skill Discovery
Paper • 2610.03872 • Published • 11 -
The Numerical Linear Algebra of Large Language Models
Paper • 2610.04631 • Published • 8 -
MEA: A Reward-Driven Multi-Agent System for Faithful Model Explanations
Paper • 2610.02480 • Published • 7 -
Closing the Context Gap: Activation Alignment for Tabular In-Context Learning
Paper • 2610.06679 • Published • 5 -
Intent Interpretation at RIC Timescales: Jev Decision Models versus Large Language Models in 6G Open RAN
Paper • 2609.23136 • Published • 6 -
The AI Theorist reveals excitonic structure in α-RuCl_3
Paper • 2610.02417 • Published • 5 -
Adapting prior-data fitted networks for tabular anomaly detection
Paper • 2610.06693 • Published • 6 -
From Evidence to Action: How Tool-Using Agents Fail
Paper • 2610.07753 • Published • 47 -
AutoSciBench: Autonomous Benchmark Generation for Evaluating Scientific Agents
Paper • 2610.05140 • Published • 46 -
MiniCorp: The Last Mile of the AI Agent Firm
Paper • 2610.05912 • Published • 39 -
DiffGate: Difficulty-Gated Teacher Guidance for On-Policy Distillation
Paper • 2610.04596 • Published • 22 -
Adaptive Latent Capacity for World Models
Paper • 2609.32921 • Published • 25 -
GUI-HARVEST: Self-Improving GUI Agents through Evidence-Driven Harness Evolution
Paper • 2610.00948 • Published • 23 -
Making LLMs Say What They Think: Measuring and Improving CoT-Interpretability Alignment
Paper • 2609.38972 • Published • 30 -
DiVeR: Decision-Critical Verifier Learning for VLA Test-Time Scaling
Paper • 2610.04933 • Published • 17 -
TimeBraid: Unifying Time Series and Language for Understanding and Forecasting
Paper • 2609.29792 • Published • 29 -
Sherpa: Teaching LLMs to Teach Adaptively
Paper • 2610.08778 • Published • 16 -
Harness-Aware Distillation for Small Language Model Agents
Paper • 2610.02858 • Published • 17 -
World Models' Last Exam in Physics
Paper • 2610.08791 • Published • 30 -
Harness Engineering for Software Engineering via Modular Executable Dev-Primitives
Paper • 2610.07832 • Published • 13 -
DAEDALUS: Bootstrapping Agent Memory from Self-Generated Tasks
Paper • 2610.08048 • Published • 18 -
We Query, Therefore We Compute: On Oracle Computation beyond the Machine, with an Application to Agents
Paper • 2610.09243 • Published • 14 -
Learning Functional Subspaces for Neural Network Compression
Paper • 2609.40127 • Published • 26 -
Learning Discriminative Geometry for Drifting Models
Paper • 2610.04703 • Published • 6 -
Judged Useless, Queried Anyway: Tool-Using Agents Rarely Turn Their Own Evidence Judgments into Stopping Decisions
Paper • 2610.06191 • Published • 4 -
A Safe Action Is Not Enough: Feasible-Future Decoding for Vision-Language-Action Policies
Paper • 2610.05166 • Published • 5 -
CheckerBench: Can Long-Horizon Agents Synthesize Static-Analysis Checkers?
Paper • 2610.07557 • Published • 74 -
JumpStart Your Policy Learning with Lessons from 160,000 Training Runs
Paper • 2609.13730 • Published • 3 -
nanoMuse: An Open-Source Personal Agent for Every Device You Own
Paper • 2610.08699 • Published • 121 -
PhysEvo: Astra Can Act, Let It
Paper • 2610.08995 • Published • 88 -
UniWAM: Unified World-Action Model
Paper • 2610.02054 • Published • 83 -
DecepEval: A Benchmark for Evaluating Deception in LLM Agents
Paper • 2610.07967 • Published • 60 -
RunningTab: Direct Workspace Interaction with Environment-Side Tabs
Paper • 2610.10444 • Published • 44 -
Mechanics of Long-Context Hybrid Models Part 1.1: From Hybrid Attention to Hybrid Position
Paper • 2610.10114 • Published • 33 -
Q-Learning with Scalar Adjoint Matching
Paper • 2610.10437 • Published • 26 -
AutoResearch at Production Scale: Failure Modes and a Multi-Agent Framework
Paper • 2609.30541 • Published • 3 -
RoboQuest: Generalist Physical Agents that Search, Inspect and Test
Paper • 2610.10388 • Published • 4 -
How corner is a corner case? Percentile control for highway scenario generation
Paper • 2610.05003 • Published • 2 -
System Switch: When Should a Fast Decision Model Stop and Think?
Paper • 2610.09683 • Published • 2 -
Synthesis Through Simulation: Generating Coherent Enterprise Data via Scalable Agent-System Interaction
Paper • 2610.10549 • Published • 8 -
Incremental Open-Ended Deep Research with Structured Harness
Paper • 2610.11566 • Published • 7 -
On-Policy Distillation Teaches New Skills but Not New Knowledge
Paper • 2610.09639 • Published • 8 -
Skill Constellations: Tracing the Supply Chain of Agent Skills on GitHub
Paper • 2610.11169 • Published • 8 -
Frozen Models, Evolving Expertise: Model-Agnostic Learning from Deployment Experience for Multimodal Medical AI
Paper • 2610.09146 • Published • 7 -
The Lattice of Transition Laws
Paper • 2610.11216 • Published • 7 -
Learning to Steer, Steering to See: Unveiling the Geometry of RLVR in Large Language Models via Trainable Vectors
Paper • 2609.34344 • Published • 7 -
AgentGarten: Code Worlds for Evolving Agents
Paper • 2610.12374 • Published • 295 -
REMORY: Learning Residual Memory for Context Compaction
Paper • 2610.11287 • Published • 64 -
Learn2Play Bench: How Well Do LLM Agents Learn from Experience in Unfamiliar Environments?
Paper • 2610.08215 • Published • 62 -
From Traces to Agentic Worlds: Agentic Language World Models for Interactive Environment Simulation
Paper • 2610.06100 • Published • 79 -
TokenRouter: Efficient Serving System for Token-Level LLM Routing
Paper • 2610.12242 • Published • 145 -
MiMo-V2.6: Scaling Reinforcement Learning Towards Self-Improvement
Paper • 2610.11959 • Published • 93 -
In-context Robot Learning Made Simple: A Democratized Recipe for Manipulation Tasks
Paper • 2609.38173 • Published • 65 -
U-Space: Uncovering When and Why Uncertainty Arises in Language Models
Paper • 2610.09087 • Published • 54 -
Memento 3: Model-Based Recursive Self-Improvement through Reflective Rulebooks
Paper • 2610.11794 • Published • 48 -
Beyond Spatio-Temporal Priors: A Generalizable Approach for Dense Correspondence Matching
Paper • 2610.12421 • Published • 21 -
MC-Sparse: Deconstructing and Closing the Dense-Sparse Attention Gap in Diffusion Transformers
Paper • 2610.06801 • Published • 46 -
Foundations of Large Language Models
Paper • 2501.09223 • Published • 51 -
SparseDecoding: Decoding-Aware Pruning for Accurate and Efficient LLM Inference
Paper • 2610.12327 • Published • 38 -
What Did the Agent Actually Do? Evidence-Grounded Oversight for Long-Horizon Agents
Paper • 2610.06406 • Published • 28 -
Embodied Turing Machines: Stateful Code for Robot Recursive Self-Improvement
Paper • 2610.12369 • Published • 28 -
SanSi: A Looped Typed Decision Model for System 1.5 Thinking
Paper • 2610.07730 • Published • 23 -
WorldGuide: Goal-Directed Video World Model for Procedural Task Execution
Paper • 2610.12459 • Published • 25 -
Opera: A Verbal Critic Framework for Long-horizon Coding Agents
Paper • 2609.33987 • Published • 19 -
Do LLMs Understand Sequential Structure? A Controlled Study of Inference and Generation
Paper • 2610.04977 • Published • 17 -
Chaos in the Text: Revealing the Modality Preference in Mixed-Modality Retrievers
Paper • 2610.11816 • Published • 23 -
A Closer Look at Agentic BBO: Benchmarking LLM Agents for Black-Box Optimization
Paper • 2610.12183 • Published • 15 -
Can AI Agents Make Open-Ended Scientific Discovery? Evidence from Station
Paper • 2610.08927 • Published • 12 -
Retrieval-Centric Deep Learning in Growing Nonparametric Neural Networks
Paper • 2610.03858 • Published • 15 -
EmbodiedCLBench: Evaluating Continual Learning for Self-Evolving Embodied Agents
Paper • 2610.13021 • Published • 76 -
Can AI Agents Learn Their Way to the Top? Evaluating Heuristic Learning in a Long-Running Game Agent Competition
Paper • 2610.12341 • Published • 49 -
Recurrent Harness for Agentic Continual Learning
Paper • 2610.12668 • Published • 27 -
SWE-Duel: Towards a Self-Scaling Arena for Coding Agents
Paper • 2610.12569 • Published • 22 -
Universal Textual Teaching for LLMs
Paper • 2610.12114 • Published • 20 -
One Model to Loop Them All: Elastic Latent Recurrence for Language Models
Paper • 2610.12582 • Published • 19 -
PreSTO: Predictive Subtree Prefetching for Fast LLM Power Sampling
Paper • 2610.13171 • Published • 15 -
Less from More: Reinforcing Sparse Video Reasoning from Dense References
Paper • 2610.10893 • Published • 13 -
The Life Cycle of a Massive Activation: Stochastic Birth, Weight-Decay-Driven Growth, and Competitive Consolidation
Paper • 2610.00423 • Published • 9 -
Distributionally Robust Mixture-of-Experts Training
Paper • 2610.07207 • Published • 10 -
Mental-Models for Multi-Agent Systems
Paper • 2610.12453 • Published • 7 -
Jeva: System One Models from First Principles
Paper • 2610.18773 • Published • 437 -
AesCode: Aesthetic Code Generation with Decoupled Cross-Modal Rewards
Paper • 2610.18860 • Published • 137 -
Harness^2: Recursive Agent Harnessing for an Open World
Paper • 2610.14224 • Published • 124 -
DoSA: Dots Sparse Attention
Paper • 2610.13990 • Published • 75 -
JevGPT: Thinking Fast and Slow in Autonomous Research
Paper • 2610.18272 • Published • 40 -
ClawLoop: Self-Improving Agents via Harness-Native Training Loop
Paper • 2610.14118 • Published • 35 -
ReTerm: Recovery Training for Self-Improving Terminal Agents
Paper • 2610.18484 • Published • 28 -
Pristine Models
Paper • 2610.14091 • Published • 30 -
Latent World Action Model for GUI Agents
Paper • 2610.14228 • Published • 24 -
EMHO: EMbodied Agent Harness Optimization via Experience Traces
Paper • 2610.08432 • Published • 10 -
AutoEnvScaling: Automating the Data Flywheel with Terminal Agents
Paper • 2610.14163 • Published • 11 -
Omni-Modal LLMs Know More Than They Say
Paper • 2610.16135 • Published • 11 -
Agent Sociodrama: Social-Dynamics-Guided Persona Recasting for LLM Agent Teams
Paper • 2610.15788 • Published • 6 -
Web$hell: A Unified Interface for Agentic Web Access
Paper • 2610.13416 • Published • 6 -
Scaling Verifiable Environments for Long-horizon Work Agents
Paper • 2610.04906 • Published • 5 -
Model-Based Online Decision Making via Generative Trajectory Planning
Paper • 2610.18677 • Published • 4 -
Can an Agent Learn to Predict Its Own Failure? Outcome-Grounded Confidence Calibration in LLM Agents
Paper • 2610.18614 • Published • 5 -
Probable but Not Valid: A Psychometric Audit of Decision Models as Synthetic Survey Respondents
Paper • 2610.14912 • Published • 3 -
TacitVLA: Reasoning Vision-Language-Action Models Without Reasoning Latency
Paper • 2610.17540 • Published • 3 -
Beyond Prompt Count: How Data Shapes Transfer in On-Policy Distillation
Paper • 2609.37377 • Published • 3 -
Words to World: Benchmarking LLMs in Interactive Physical Environments
Paper • 2610.15228 • Published • 3 -
Thinking Inertia: LLMs Keep Thinking When Told Not To
Paper • 2610.11765 • Published • 244 -
AgentSWE: Can Coding Agents Build the Agent You Actually Want?
Paper • 2610.13361 • Published • 92 -
Learning More by Training Less: Uncertainty-Guided Progressive Contraction for Efficient Reasoning Post-Training
Paper • 2610.12933 • Published • 14 -
ARCS: Towards Precise Text-to-SQL via Structured Disambiguation
Paper • 2610.09396 • Published • 13 -
Visual Grounding Safety in Vision-Language Models
Paper • 2610.05637 • Published • 12 -
Branching On-Policy Distillation
Paper • 2610.13481 • Published • 11 -
HyperThink: Text-to-Parameter Hypernetworks for Efficient Reasoning
Paper • 2610.03039 • Published • 8 -
Same Output, Different Gold: Measuring How Reference Choice Moves a Multilingual Benchmark Score
Paper • 2610.03825 • Published • 6 -
Agent-Reflex: Teaching Language Agents to Act Reflectively
Paper • 2610.17953 • Published • 216 -
Self-Distillation with Verbal Regularization
Paper • 2610.16354 • Published • 34 -
LeWAM: End-to-End World and Action Modeling with JEPAs
Paper • 2610.12818 • Published • 32 -
PsyBer-Agent: A Psychology-Driven User Agent for Recommender Systems
Paper • 2610.13989 • Published • 9 -
StartLux-Decision: AutoThink Decision Models with Reinforcement Learning for Calibrated Deliberative Decisions
Paper • 2610.17309 • Published • 81 -
Many Right Answers, One Training Example: Removing Responder Noise from GRPO for LLM Recommenders
Paper • 2610.17771 • Published • 79 -
APEX-Harm: Evaluating Agent Safety in Shared Professional Workspaces
Paper • 2610.13623 • Published • 26 -
ProgAgent: Learning Tool Orchestration through Programmatic Training
Paper • 2610.14774 • Published • 19 -
What Makes a Social Agent's Interactions Effective? Learning from Verifiable Outcomes in Paired Worlds
Paper • 2610.20195 • Published • 17 -
Next-Concept Modeling: Fast, Parallel Coarse-to-Fine Text Generation
Paper • 2610.14162 • Published • 17 -
LLM-to-Flow Distillation: Teaching Natural Language Goals to Scientific Generators
Paper • 2610.17246 • Published • 15 -
ProbabilityBench: Model Evaluation and Calibration with Exact Ground-Truth Uncertainty
Paper • 2610.17016 • Published • 14 -
Continuum Attention
Paper • 2610.20586 • Published • 10 -
Choosing LLMs for Strategic Agent Interactions
Paper • 2610.18354 • Published • 8 -
AgentSafeEvo: Adversarial Co-Evolution for LLM Agent Safety
Paper • 2610.19422 • Published • 7 -
Cortex: Consolidating Context into Parametric Memory
Paper • 2610.20399 • Published • 73 -
ActionOPD: Distilling Decisions, Not Tokens
Paper • 2610.19820 • Published • 64 -
Doc2Harness: Benchmarking LLMs on Long-Horizon Harness Engineering of Coding Agents
Paper • 2610.20344 • Published • 62 -
More Methods, Better Agents? A Reality Check on Self-Distillation in Agentic Reinforcement Learning
Paper • 2610.19789 • Published • 60 -
Seeing, Maintaining, and Learning: The Evolution of Multimodal Agent Memory
Paper • 2610.19366 • Published • 56 -
Towards Agentic Robots: Model, Data, Environment, and Harness
Paper • 2610.20442 • Published • 43 -
Hybrid Reasoning Agents
Paper • 2610.18861 • Published • 44 -
CLASHarness: Benchmarking Multi-Agent Conflict in Computer Use via a Controllable and Verifiable Generation Pipeline
Paper • 2610.19128 • Published • 36 -
The Thinking Leakage in Reinforcement Learning for LLMs
Paper • 2610.14270 • Published • 33 -
Long-Horizon RL Needs Muon
Paper • 2610.20256 • Published • 30 -
Apertus 1.5: From Open Text Models to Open Multimodal Reasoners
Paper • 2610.15542 • Published • 53