Xu Han

InfLLM-V2: Dense-Sparse Switchable Attention for Seamless Short-to-Long Adaptation

Process Reinforcement through Implicit Rewards

AUTOTRITON: Automatic Triton Programming with Reinforcement Learning in LLMs

StateX: Enhancing RNN Recall via Post-training State Expansion

Sparsing Law: Towards Large Language Models with Greater Activation Sparsity

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

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

Stuffed Mamba: Oversized States Lead to the Inability to Forget

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

Locret: Enhancing Eviction in Long-Context LLM Inference with Trained Retaining Heads

Sparsing Law: Towards Large Language Models with Greater Activation Sparsity

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