Jun Zhao

Omni-Reward: Towards Generalist Omni-Modal Reward Modeling with Free-Form Preferences

MMR-Life: Piecing Together Real-life Scenes for Multimodal Multi-image Reasoning

MMR-V: What's Left Unsaid? A Benchmark for Multimodal Deep Reasoning in Videos

Fixing the Broken Compass: Diagnosing and Improving Inference-Time Reward Modeling

DAComp: Benchmarking Data Agents across the Full Data Intelligence Lifecycle

Pixels Lie, Code Doesn't: Thinking with Visual Programming for ''Seemingly Impossible'' Multimodal Agentic Reasoning Tasks

Reinforcing Agentic Search Via Reward Density Optimization

Amplify Adjacent Token Differences: Enhancing Long Chain-of-Thought Reasoning with Shift-FFN

Reinforced Internal-External Knowledge Synergistic Reasoning for Efficient Adaptive Search Agent

Combination Generalization of Capability-Specific Neurons in LLMs

Towards Atoms of Large Language Models

Towards Agentic Self-Learning LLMs in Search Environment

Beyond Hard and Soft: Hybrid Context Compression for Balancing Local and Global Information Retention

Imitation Game: Toward Comprehensive Evaluation on Personalized Role-Playing on Social Media

WebPlanner: Task Planning with Autonomous Experience Exploration and Utilization for Real World Multimodal Web Agents

Can LLM Event Prediction Be Reliable? Closing Gaps in Causal Quantification and Probabilistic Consistency

Dynamic Parametric Retrieval Augmented Generation

BDQ: Bidirectional Diagonal Quantization for LLMs

Probabilistic Uncertain Reward Model

Effective Interaction between Quantization and Low-Rank Decomposition based on LLMs

The Zero-Step Thinking: An Empirical Study of Mode Selection as a Harder Early Exit Problem in Reasoning Models

Knowledge Localization: Mission Not Accomplished? Enter Query Localization!

RULE: Reinforcement UnLEarning Achieves Forget-retain Pareto Optimality

Fixing the Broken Compass: Diagnosing and Improving Inference-Time Reward Modeling

MIRAGE: Evaluating and Explaining Inductive Reasoning Process in Language Models

Capability Localization: Capabilities Can be Localized rather than Individual Knowledge

M2Edit: Locate and Edit Multi-Granularity Knowledge in Multimodal Large Language Model