EoRA: Training-free Compensation for Compressed LLM with Eigenspace Low-Rank Approximation

Open in new window