Fetching the paper…

Compressing Pre-trained Transformers via Low-Bit NxM Sparsity for Natural Language Understanding · Around