Maosong Sun

From f(x) and g(x) to f(g(x)): LLMs Learn New Skills in RL by Composing Old Ones

KG-Infused RAG: Augmenting Corpus-Based RAG with External Knowledge Graphs

CPMöbius: Iterative Coach–Player Reasoning for Data-Free Reinforcement Learning

Test-Time Exploration in Unknown Environments

Process Reinforcement through Implicit Rewards

Exploration for Building Next-Generation Foundation MLLMs via Self-Learning

Evidence-Guided Multi-Image Reasoning in Visual Retrieval-Augmented Generation

LLaVA-UHD v3: Progressive Visual Compression for Efficient Naive-Resolution Encoding in MLLMs

KARE-RAG: Knowledge-Aware Refinement and Enhancement for RAG

Listens like Mel: Boosting Latent Audio Diffusion with Channel Locality

Query Routing over Multimodal Knowledge Bases for Retrieval-Augmented Reasoning

AUTOTRITON: Automatic Triton Programming with Reinforcement Learning in LLMs

RLPR: Extrapolating RLVR to General Domains without Verifiers

Musical Score Understanding Benchmark: Evaluating Large Language Models’ Comprehension of Complete Musical Scores

StateX: Enhancing RNN Recall via Post-training State Expansion

EMind: A Foundation Model for Multi-task Electromagnetic Signals Understanding

Diversity-aware Training for Test-time Scaling

Reflective Reinforcement Tool Learning

Sparsing Law: Towards Large Language Models with Greater Activation Sparsity

A*-Thought: Efficient Reasoning via Bidirectional Compression for Low-Resource Settings

Internet of Agents: Weaving a Web of Heterogeneous Agents for Collaborative Intelligence

The Overthinker's DIET: Cutting Token Calories with DIfficulty-AwarE Training

BlockFFN: Towards End-Side Acceleration-Friendly Mixture-of-Experts with Chunk-Level Activation Sparsity

Scaling Large Language Model-based Multi-Agent Collaboration

AIR: A Systematic Analysis of Annotations, Instructions, and Response Pairs in Preference Dataset

ParamMute: Suppressing Knowledge-Critical FFNs for Faithful Retrieval-Augmented Generation

Advancing LLM Reasoning Generalists with Preference Trees

Multi-Agent Collaboration via Evolving Orchestration

WorkflowLLM: Enhancing Workflow Orchestration Capability of Large Language Models

Stuffed Mamba: Oversized States Lead to the Inability to Forget

A Multi-Power Law for Loss Curve Prediction Across Learning Rate Schedules

VisRAG: Vision-based Retrieval-augmented Generation on Multi-modality Documents

RAG-DDR: Optimizing Retrieval-Augmented Generation Using Differentiable Data Rewards

StreamingBench: Assessing the Gap for MLLMs to Achieve Streaming Video Understanding

Rational Decision-Making Agent with Learning Internal Utility Judgment

Optima: Optimizing Effectiveness and Efficiency for LLM-Based Multi-Agent System

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

Sparsing Law: Towards Large Language Models with Greater Activation Sparsity

Selecting Influential Samples for Long Context Alignment via Homologous Models’ Guidance and Contextual Awareness Measurement

Improving Zero-Shot Generalization of Instruction Tuning by Data Arrangement

Stuffed Mamba: State Collapse and State Capacity of RNN-Based Long-Context Modeling