Yankai Lin

ViPER: Empowering the Self-Evolution of Visual Perception Abilities in Vision-Language Models

LaSeR: Reinforcement Learning with Last-Token Self-Rewarding

DeepCritic: Deliberate Critique with Large Language Models

DeepThinkVLA: Enhancing Reasoning Capability of Vision-Language-Action Models

Union-of-Experts: Experts in Mixture-of-Experts are Secretly Routers

Test-Time Exploration in Unknown Environments

LLaDA 1.5: Variance-Reduced Preference Optimization for Large Language Diffusion Models

Reflective Reinforcement Tool Learning

Large Language Diffusion Models

Uncertainty and Influence aware Reward Model Refinement for Reinforcement Learning from Human Feedback

Generalizing Experience for Language Agents with Hierarchical MetaFlows

Super(ficial)-alignment: Strong Models May Deceive Weak Models in Weak-to-Strong Generalization

Advancing LLM Reasoning Generalists with Preference Trees

Towards Thinking-Optimal Scaling of Test-Time Compute for LLM Reasoning

Learning to Focus: Causal Attention Distillation via Gradient‐Guided Token Pruning

WorkflowLLM: Enhancing Workflow Orchestration Capability of Large Language Models

Rational Decision-Making Agent with Learning Internal Utility Judgment

Proactive Agent: Shifting LLM Agents from Reactive Responses to Active Assistance

Large Language Diffusion Models

Large-Scale Dynamic Graph Generation via LLM-based Agent Simulation

Learning Evolving Tools for Large Language Models