Enabling Lightweight Fine-tuning for Pre-trained Language Model Compression based on Matrix Product Operators

8279

Related Entities