Liyuan Liu

DenseMixer: Improving MoE Post-Training with Precise Router Gradient

Mixture of Inputs: Text Generation Beyond Discrete Token Sampling

Reinforcement Learning for Reasoning in Large Language Models with One Training Example

Routing Mamba: Scaling State Space Models with Mixture-of-Experts Projection

Decoder-Hybrid-Decoder Architecture for Efficient Reasoning with Long Generation

Training Language Models to Generate Quality Code with Program Analysis Feedback

Vector-ICL: In-context Learning with Continuous Vector Representations

LoRC: Low-Rank Compression for LLMs KV Cache with a Progressive Compression Strategy

Model Tells Itself Where to Attend: Steerable Prompting for Reliable Reading Comprehension of LLM