Youtu-LLM: Unlocking the Native Agentic Potential for Lightweight Large Language Models
Paper
•
2512.24618
•
Published
•
149
2026
-
-
Let It Flow: Agentic Crafting on Rock and Roll, Building the ROME Model within an Open Agentic Learning Ecosystem
Paper • 2512.24873 • Published • 104 -
AI Meets Brain: Memory Systems from Cognitive Neuroscience to Autonomous Agents
Paper • 2512.23343 • Published • 29 -
Figure It Out: Improving the Frontier of Reasoning with Active Visual Thinking
Paper • 2512.24297 • Published • 6 -
Valori: A Deterministic Memory Substrate for AI Systems
Paper • 2512.22280 • Published • 5 -
Improving Multi-step RAG with Hypergraph-based Memory for Long-Context Complex Relational Modeling
Paper • 2512.23959 • Published • 112 -
Youtu-Agent: Scaling Agent Productivity with Automated Generation and Hybrid Policy Optimization
Paper • 2512.24615 • Published • 119 -
Nested Learning: The Illusion of Deep Learning Architectures
Paper • 2512.24695 • Published • 43 -
SenseNova-MARS: Empowering Multimodal Agentic Reasoning and Search via Reinforcement Learning
Paper • 2512.24330 • Published • 35 -
Fast-weight Product Key Memory
Paper • 2601.00671 • Published • 6 -
SimpleMem: Efficient Lifelong Memory for LLM Agents
Paper • 2601.02553 • Published • 36 -
Falcon-H1R: Pushing the Reasoning Frontiers with a Hybrid Model for Efficient Test-Time Scaling
Paper • 2601.02346 • Published • 26 -
OpenNovelty: An LLM-powered Agentic System for Verifiable Scholarly Novelty Assessment
Paper • 2601.01576 • Published • 18 -
UniCorn: Towards Self-Improving Unified Multimodal Models through Self-Generated Supervision
Paper • 2601.03193 • Published • 47 -
NitroGen: An Open Foundation Model for Generalist Gaming Agents
Paper • 2601.02427 • Published • 44 -
MindWatcher: Toward Smarter Multimodal Tool-Integrated Reasoning
Paper • 2512.23412 • Published • 40 -
Token-Level LLM Collaboration via FusionRoute
Paper • 2601.05106 • Published • 40 -
AT^2PO: Agentic Turn-based Policy Optimization via Tree Search
Paper • 2601.04767 • Published • 28 -
Agent-as-a-Judge
Paper • 2601.05111 • Published • 19 -
WebGym: Scaling Training Environments for Visual Web Agents with Realistic Tasks
Paper • 2601.02439 • Published • 16 -
The Illusion of Specialization: Unveiling the Domain-Invariant "Standing Committee" in Mixture-of-Experts Models
Paper • 2601.03425 • Published • 16 -
Scaling Behavior Cloning Improves Causal Reasoning: An Open Model for Real-Time Video Game Playing
Paper • 2601.04575 • Published • 9 -
DocDancer: Towards Agentic Document-Grounded Information Seeking
Paper • 2601.05163 • Published • 5 -
Entropy-Adaptive Fine-Tuning: Resolving Confident Conflicts to Mitigate Forgetting
Paper • 2601.02151 • Published • 108 -
AgentDevel: Reframing Self-Evolving LLM Agents as Release Engineering
Paper • 2601.04620 • Published • 3 -
Evolving Programmatic Skill Networks
Paper • 2601.03509 • Published • 84 -
Atlas: Orchestrating Heterogeneous Models and Tools for Multi-Domain Complex Reasoning
Paper • 2601.03872 • Published • 42 -
Thinking with Map: Reinforced Parallel Map-Augmented Agent for Geolocalization
Paper • 2601.05432 • Published • 166 -
The Molecular Structure of Thought: Mapping the Topology of Long Chain-of-Thought Reasoning
Paper • 2601.06002 • Published • 52 -
Agentic Rubrics as Contextual Verifiers for SWE Agents
Paper • 2601.04171 • Published • 11 -
Chaining the Evidence: Robust Reinforcement Learning for Deep Search Agents with Citation-Aware Rubric Rewards
Paper • 2601.06021 • Published • 45 -
MDAgent2: Large Language Model for Code Generation and Knowledge Q&A in Molecular Dynamics
Paper • 2601.02075 • Published • 8 -
EnvScaler: Scaling Tool-Interactive Environments for LLM Agent via Programmatic Synthesis
Paper • 2601.05808 • Published • 36 -
Why LLMs Aren't Scientists Yet: Lessons from Four Autonomous Research Attempts
Paper • 2601.03315 • Published • 6 -
AgentOCR: Reimagining Agent History via Optical Self-Compression
Paper • 2601.04786 • Published • 29 -
MAGMA: A Multi-Graph based Agentic Memory Architecture for AI Agents
Paper • 2601.03236 • Published • 3 -
Can We Predict Before Executing Machine Learning Agents?
Paper • 2601.05930 • Published • 26 -
GDPO: Group reward-Decoupled Normalization Policy Optimization for Multi-reward RL Optimization
Paper • 2601.05242 • Published • 222 -
An Empirical Study on Preference Tuning Generalization and Diversity Under Domain Shift
Paper • 2601.05882 • Published • 20 -
Illusions of Confidence? Diagnosing LLM Truthfulness via Neighborhood Consistency
Paper • 2601.05905 • Published • 18 -
SmartSearch: Process Reward-Guided Query Refinement for Search Agents
Paper • 2601.04888 • Published • 10 -
Over-Searching in Search-Augmented Large Language Models
Paper • 2601.05503 • Published • 6 -
DR-LoRA: Dynamic Rank LoRA for Mixture-of-Experts Adaptation
Paper • 2601.04823 • Published • 6 -
Memory Matters More: Event-Centric Memory as a Logic Map for Agent Searching and Reasoning
Paper • 2601.04726 • Published • 6 -
TCAndon-Router: Adaptive Reasoning Router for Multi-Agent Collaboration
Paper • 2601.04544 • Published • 6 -
IIB-LPO: Latent Policy Optimization via Iterative Information Bottleneck
Paper • 2601.05870 • Published • 3 -
Distilling Feedback into Memory-as-a-Tool
Paper • 2601.05960 • Published • 2 -
BabyVision: Visual Reasoning Beyond Language
Paper • 2601.06521 • Published • 196 -
PaCoRe: Learning to Scale Test-Time Compute with Parallel Coordinated Reasoning
Paper • 2601.05593 • Published • 83 -
Lost in the Noise: How Reasoning Models Fail with Contextual Distractors
Paper • 2601.07226 • Published • 32 -
Dr. Zero: Self-Evolving Search Agents without Training Data
Paper • 2601.07055 • Published • 20 -
OS-Symphony: A Holistic Framework for Robust and Generalist Computer-Using Agent
Paper • 2601.07779 • Published • 28 -
Controllable Memory Usage: Balancing Anchoring and Innovation in Long-Term Human-Agent Interaction
Paper • 2601.05107 • Published • 24 -
ET-Agent: Incentivizing Effective Tool-Integrated Reasoning Agent via Behavior Calibration
Paper • 2601.06860 • Published • 16 -
MegaFlow: Large-Scale Distributed Orchestration System for the Agentic Era
Paper • 2601.07526 • Published • 23 -
Forest Before Trees: Latent Superposition for Efficient Visual Reasoning
Paper • 2601.06803 • Published • 10 -
TourPlanner: A Competitive Consensus Framework with Constraint-Gated Reinforcement Learning for Travel Planning
Paper • 2601.04698 • Published • 10 -
How Do Large Language Models Learn Concepts During Continual Pre-Training?
Paper • 2601.03570 • Published • 4 -
OpenTinker: Separating Concerns in Agentic Reinforcement Learning
Paper • 2601.07376 • Published • 6 -
ShowUI-Aloha: Human-Taught GUI Agent
Paper • 2601.07181 • Published • 3 -
Are LLM Decisions Faithful to Verbal Confidence?
Paper • 2601.07767 • Published • 4 -
Structured Episodic Event Memory
Paper • 2601.06411 • Published • 4 -
Artificial Entanglement in the Fine-Tuning of Large Language Models
Paper • 2601.06788 • Published • 3 -
User-Oriented Multi-Turn Dialogue Generation with Tool Use at scale
Paper • 2601.08225 • Published • 52 -
ArenaRL: Scaling RL for Open-Ended Agents via Tournament-based Relative Ranking
Paper • 2601.06487 • Published • 52 -
On the Non-decoupling of Supervised Fine-tuning and Reinforcement Learning in Post-training
Paper • 2601.07389 • Published • 2 -
MemoBrain: Executive Memory as an Agentic Brain for Reasoning
Paper • 2601.08079 • Published • 37 -
MemGovern: Enhancing Code Agents through Learning from Governed Human Experiences
Paper • 2601.06789 • Published • 78 -
The Confidence Dichotomy: Analyzing and Mitigating Miscalibration in Tool-Use Agents
Paper • 2601.07264 • Published • 24 -
Parallel Context-of-Experts Decoding for Retrieval Augmented Generation
Paper • 2601.08670 • Published • 19 -
Aligning Text, Code, and Vision: A Multi-Objective Reinforcement Learning Framework for Text-to-Visualization
Paper • 2601.04582 • Published • 10 -
JudgeRLVR: Judge First, Generate Second for Efficient Reasoning
Paper • 2601.08468 • Published • 6 -
EpiCaR: Knowing What You Don't Know Matters for Better Reasoning in LLMs
Paper • 2601.06786 • Published • 6 -
Controlled Self-Evolution for Algorithmic Code Optimization
Paper • 2601.07348 • Published • 114 -
MAXS: Meta-Adaptive Exploration with LLM Agents
Paper • 2601.09259 • Published • 95 -
EvoFSM: Controllable Self-Evolution for Deep Research with Finite State Machines
Paper • 2601.09465 • Published • 41 -
OpenDecoder: Open Large Language Model Decoding to Incorporate Document Quality in RAG
Paper • 2601.09028 • Published • 33 -
ExpSeek: Self-Triggered Experience Seeking for Web Agents
Paper • 2601.08605 • Published • 16 -
Imagine-then-Plan: Agent Learning from Adaptive Lookahead with World Models
Paper • 2601.08955 • Published • 13 -
No More Stale Feedback: Co-Evolving Critics for Open-World Agent Learning
Paper • 2601.06794 • Published • 4 -
The AI Hippocampus: How Far are We From Human Memory?
Paper • 2601.09113 • Published • 5 -
DPWriter: Reinforcement Learning with Diverse Planning Branching for Creative Writing
Paper • 2601.09609 • Published • 3 -
Omni-R1: Towards the Unified Generative Paradigm for Multimodal Reasoning
Paper • 2601.09536 • Published • 5 -
SCALER:Synthetic Scalable Adaptive Learning Environment for Reasoning
Paper • 2601.04809 • Published • 3 -
Rewarding the Rare: Uniqueness-Aware RL for Creative Problem Solving in LLMs
Paper • 2601.08763 • Published • 147 -
Collaborative Multi-Agent Test-Time Reinforcement Learning for Reasoning
Paper • 2601.09667 • Published • 87 -
Beyond Static Tools: Test-Time Tool Evolution for Scientific Reasoning
Paper • 2601.07641 • Published • 46 -
Toward Ultra-Long-Horizon Agentic Science: Cognitive Accumulation for Machine Learning Engineering
Paper • 2601.10402 • Published • 36 -
MatchTIR: Fine-Grained Supervision for Tool-Integrated Reasoning via Bipartite Matching
Paper • 2601.10712 • Published • 24 -
LaViT: Aligning Latent Visual Thoughts for Multi-modal Reasoning
Paper • 2601.10129 • Published • 11 -
PACEvolve: Enabling Long-Horizon Progress-Aware Consistent Evolution
Paper • 2601.10657 • Published • 20 -
LSRIF: Logic-Structured Reinforcement Learning for Instruction Following
Paper • 2601.06431 • Published • 12 -
PRL: Process Reward Learning Improves LLMs' Reasoning Ability and Broadens the Reasoning Boundary
Paper • 2601.10201 • Published • 8 -
Agent Skills in the Wild: An Empirical Study of Security Vulnerabilities at Scale
Paper • 2601.10338 • Published • 5 -
Memory Bank Compression for Continual Adaptation of Large Language Models
Paper • 2601.00756 • Published • 2 -
Distribution-Aligned Sequence Distillation for Superior Long-CoT Reasoning
Paper • 2601.09088 • Published • 62 -
Your Group-Relative Advantage Is Biased
Paper • 2601.08521 • Published • 150 -
The Poisoned Apple Effect: Strategic Manipulation of Mediated Markets via Technology Expansion of AI Agents
Paper • 2601.11496 • Published • 47 -
Unlocking Implicit Experience: Synthesizing Tool-Use Trajectories from Text
Paper • 2601.10355 • Published • 39 -
BAPO: Boundary-Aware Policy Optimization for Reliable Agentic Search
Paper • 2601.11037 • Published • 17 -
ProFit: Leveraging High-Value Signals in SFT via Probability-Guided Token Selection
Paper • 2601.09195 • Published • 15 -
Reasoning Models Generate Societies of Thought
Paper • 2601.10825 • Published • 14 -
PersonalAlign: Hierarchical Implicit Intent Alignment for Personalized GUI Agent with Long-Term User-Centric Records
Paper • 2601.09636 • Published • 8 -
Language of Thought Shapes Output Diversity in Large Language Models
Paper • 2601.11227 • Published • 9 -
Multiplex Thinking: Reasoning via Token-wise Branch-and-Merge
Paper • 2601.08808 • Published • 39 -
NAACL: Noise-AwAre Verbal Confidence Calibration for LLMs in RAG Systems
Paper • 2601.11004 • Published • 30 -
Spurious Rewards Paradox: Mechanistically Understanding How RLVR Activates Memorization Shortcuts in LLMs
Paper • 2601.11061 • Published • 7 -
YaPO: Learnable Sparse Activation Steering Vectors for Domain Adaptation
Paper • 2601.08441 • Published • 7 -
CLARE: Continual Learning for Vision-Language-Action Models via Autonomous Adapter Routing and Expansion
Paper • 2601.09512 • Published • 4 -
Think3D: Thinking with Space for Spatial Reasoning
Paper • 2601.13029 • Published • 47 -
Toward Efficient Agents: Memory, Tool learning, and Planning
Paper • 2601.14192 • Published • 54 -
DARC: Decoupled Asymmetric Reasoning Curriculum for LLM Evolution
Paper • 2601.13761 • Published • 16 -
Aligning Agentic World Models via Knowledgeable Experience Learning
Paper • 2601.13247 • Published • 15 -
Agentic-R: Learning to Retrieve for Agentic Search
Paper • 2601.11888 • Published • 19 -
Which Reasoning Trajectories Teach Students to Reason Better? A Simple Metric of Informative Alignment
Paper • 2601.14249 • Published • 10 -
InT: Self-Proposed Interventions Enable Credit Assignment in LLM Reasoning
Paper • 2601.14209 • Published • 6 -
Uncertainty-Aware Gradient Signal-to-Noise Data Selection for Instruction Tuning
Paper • 2601.13697 • Published • 4 -
Agentic Reasoning for Large Language Models
Paper • 2601.12538 • Published • 194 -
Paper2Rebuttal: A Multi-Agent Framework for Transparent Author Response Assistance
Paper • 2601.14171 • Published • 48 -
Behavior Knowledge Merge in Reinforced Agentic Models
Paper • 2601.13572 • Published • 24 -
Render-of-Thought: Rendering Textual Chain-of-Thought as Images for Visual Latent Reasoning
Paper • 2601.14750 • Published • 17 -
Numina-Lean-Agent: An Open and General Agentic Reasoning System for Formal Mathematics
Paper • 2601.14027 • Published • 12 -
Lost in the Prompt Order: Revealing the Limitations of Causal Attention in Language Models
Paper • 2601.14152 • Published • 5 -
The Responsibility Vacuum: Organizational Failure in Scaled Agent Systems
Paper • 2601.15059 • Published • 3 -
Facilitating Proactive and Reactive Guidance for Decision Making on the Web: A Design Probe with WebSeek
Paper • 2601.15100 • Published • 3 -
EvoCUA: Evolving Computer Use Agents via Learning from Scalable Synthetic Experience
Paper • 2601.15876 • Published • 89 -
LLM-in-Sandbox Elicits General Agentic Intelligence
Paper • 2601.16206 • Published • 84 -
PROGRESSLM: Towards Progress Reasoning in Vision-Language Models
Paper • 2601.15224 • Published • 12 -
Agentic Uncertainty Quantification
Paper • 2601.15703 • Published • 8 -
Agentic Confidence Calibration
Paper • 2601.15778 • Published • 5 -
From Passive Metric to Active Signal: The Evolving Role of Uncertainty Quantification in Large Language Models
Paper • 2601.15690 • Published • 4 -
SWE-Pruner: Self-Adaptive Context Pruning for Coding Agents
Paper • 2601.16746 • Published • 89 -
VisGym: Diverse, Customizable, Scalable Environments for Multimodal Agents
Paper • 2601.16973 • Published • 40 -
Inference-Time Scaling of Verification: Self-Evolving Deep Research Agents via Test-Time Rubric-Guided Verification
Paper • 2601.15808 • Published • 20 -
Endless Terminals: Scaling RL Environments for Terminal Agents
Paper • 2601.16443 • Published • 16 -
Dancing in Chains: Strategic Persuasion in Academic Rebuttal via Theory of Mind
Paper • 2601.15715 • Published • 13 -
ChartVerse: Scaling Chart Reasoning via Reliable Programmatic Synthesis from Scratch
Paper • 2601.13606 • Published • 11 -
MeepleLM: A Virtual Playtester Simulating Diverse Subjective Experiences
Paper • 2601.07251 • Published • 11 -
Knowledge is Not Enough: Injecting RL Skills for Continual Adaptation
Paper • 2601.11258 • Published • 8 -
Guidelines to Prompt Large Language Models for Code Generation: An Empirical Characterization
Paper • 2601.13118 • Published • 1 -
daVinci-Dev: Agent-native Mid-training for Software Engineering
Paper • 2601.18418 • Published • 124 -
Teaching Models to Teach Themselves: Reasoning at the Edge of Learnability
Paper • 2601.18778 • Published • 40 -
Paying Less Generalization Tax: A Cross-Domain Generalization Study of RL Training for LLM Agents
Paper • 2601.18217 • Published • 11 -
DRPG (Decompose, Retrieve, Plan, Generate): An Agentic Framework for Academic Rebuttal
Paper • 2601.18081 • Published • 7 -
Least-Loaded Expert Parallelism: Load Balancing An Imbalanced Mixture-of-Experts
Paper • 2601.17111 • Published • 5 -
Agentic Search in the Wild: Intents and Trajectory Dynamics from 14M+ Real Search Requests
Paper • 2601.17617 • Published • 2 -
RouteMoA: Dynamic Routing without Pre-Inference Boosts Efficient Mixture-of-Agents
Paper • 2601.18130 • Published • 1 -
AdaReasoner: Dynamic Tool Orchestration for Iterative Visual Reasoning
Paper • 2601.18631 • Published • 47 -
Self-Distillation Enables Continual Learning
Paper • 2601.19897 • Published • 24 -
Harder Is Better: Boosting Mathematical Reasoning via Difficulty-Aware GRPO and Multi-Aspect Question Reformulation
Paper • 2601.20614 • Published • 116 -
Innovator-VL: A Multimodal Large Language Model for Scientific Discovery
Paper • 2601.19325 • Published • 78 -
Reinforcement Learning via Self-Distillation
Paper • 2601.20802 • Published • 39 -
Spark: Strategic Policy-Aware Exploration via Dynamic Branching for Long-Horizon Agentic Learning
Paper • 2601.20209 • Published • 22 -
Linear representations in language models can change dramatically over a conversation
Paper • 2601.20834 • Published • 21 -
SERA: Soft-Verified Efficient Repository Agents
Paper • 2601.20789 • Published • 11 -
Group Distributionally Robust Optimization-Driven Reinforcement Learning for LLM Reasoning
Paper • 2601.19280 • Published • 9 -
OmegaUse: Building a General-Purpose GUI Agent for Autonomous Task Execution
Paper • 2601.20380 • Published • 8 -
How AI Impacts Skill Formation
Paper • 2601.20245 • Published • 8 -
VERGE: Formal Refinement and Guidance Engine for Verifiable LLM Reasoning
Paper • 2601.20055 • Published • 6 -
Training Reasoning Models on Saturated Problems via Failure-Prefix Conditioning
Paper • 2601.20829 • Published • 5 -
Idea2Story: An Automated Pipeline for Transforming Research Concepts into Complete Scientific Narratives
Paper • 2601.20833 • Published • 173 -
Scaling Embeddings Outperforms Scaling Experts in Language Models
Paper • 2601.21204 • Published • 98 -
ConceptMoE: Adaptive Token-to-Concept Compression for Implicit Compute Allocation
Paper • 2601.21420 • Published • 42 -
Exploring Reasoning Reward Model for Agents
Paper • 2601.22154 • Published • 22 -
Language-based Trial and Error Falls Behind in the Era of Experience
Paper • 2601.21754 • Published • 16 -
Self-Improving Pretraining: using post-trained models to pretrain better models
Paper • 2601.21343 • Published • 16 -
Scalable Power Sampling: Unlocking Efficient, Training-Free Reasoning for LLMs via Distribution Sharpening
Paper • 2601.21590 • Published • 12 -
Beyond Imitation: Reinforcement Learning for Active Latent Planning
Paper • 2601.21598 • Published • 9 -
DeepSearchQA: Bridging the Comprehensiveness Gap for Deep Research Agents
Paper • 2601.20975 • Published • 9 -
VTC-R1: Vision-Text Compression for Efficient Long-Context Reasoning
Paper • 2601.22069 • Published • 7 -
Reinforcement Learning from Meta-Evaluation: Aligning Language Models Without Ground-Truth Labels
Paper • 2601.21268 • Published • 4 -
BMAM: Brain-inspired Multi-Agent Memory Framework
Paper • 2601.20465 • Published • 4 -
FROST: Filtering Reasoning Outliers with Attention for Efficient Reasoning
Paper • 2601.19001 • Published • 4 -
WebArbiter: A Principle-Guided Reasoning Process Reward Model for Web Agents
Paper • 2601.21872 • Published