Less is More: Task-aware Layer-wise Distillation for Language Model Compression

Publication
The 40th International Conference on Machine Learning (ICML), 2023