Lili Mou

Cactus: Accelerating Auto-Regressive Decoding with Constrained Acceptance Speculative Sampling

Exploring Minimum Bayes Risk Decoding for Text-to-SQL Ensemble

BRIM: Block-wise Return Induction Method for Sequence Knowledge Distillation

Reward Model Boosting for RLHF

TokMem: One-Token Procedural Memory for Large Language Models

LoraQuant: Mixed-Precision Quantization of LoRA to Ultra-Low Bits

ULPT: Prompt Tuning with Ultra-Low-Dimensional Optimization

NeuZip: Memory-Efficient Training and Inference with Dynamic Compression of Neural Networks