Maksim Khadkevich

Latest

• EoRA: Fine-tuning-free Compensation for Compressed LLM with Eigenspace Low-Rank Approximation