Weight Distillation: Transferring the Knowledge in Neural Network Parameters
arXiv:2009.09152
Abstract
Knowledge distillation has been proven to be effective in model acceleration and compression. It allows a small network to learn to generalize in the same way as a large network. Recent successes in pre-training suggest the effectiveness of transferring model parameters. Inspired by this, we investigate methods of model acceleration and compression in another line of research. We propose Weight Distillation to transfer the knowledge in the large network parameters through a parameter generator. Our experiments on WMT16 En-Ro, NIST12 Zh-En, and WMT14 En-De machine translation tasks show that weight distillation can train a small network that is 1.88~2.94x faster than the large network but with competitive performance. With the same sized small network, weight distillation can outperform knowledge distillation by 0.51~1.82 BLEU points.
arXiv commentaccepted by ACL2021
References in corpus (9)
- Distilling the Knowledge in a Neural Network
- Mixed Precision Training
- Born Again Neural Networks
- Ensemble Distillation for Neural Machine Translation
- Accelerating Neural Transformer via an Average Attention Network
- Explaining Sequence-Level Knowledge Distillation as Data-Augmentation for Neural Machine Translation
- Self-Knowledge Distillation in Natural Language Processing
- Neural Machine Translation with Joint Representation
- Learning Light-Weight Translation Models from Deep Transformer