CALR: Corrective Adaptive Low-Rank Decomposition for Efficient Large Language Model Layer Compression

Open in new window