Contrastive Distillation on Intermediate Representations for Language Model Compression

Siqi Sun, Zhe Gan, Yuwei Fang, Yu Cheng, Shuohang Wang, Jingjing Liu


Abstract
Existing language model compression methods mostly use a simple L_2 loss to distill knowledge in the intermediate representations of a large BERT model to a smaller one. Although widely used, this objective by design assumes that all the dimensions of hidden representations are independent, failing to capture important structural knowledge in the intermediate layers of the teacher network. To achieve better distillation efficacy, we propose Contrastive Distillation on Intermediate Representations (CoDIR), a principled knowledge distillation framework where the student is trained to distill knowledge through intermediate layers of the teacher via a contrastive objective. By learning to distinguish positive sample from a large set of negative samples, CoDIR facilitates the student’s exploitation of rich information in teacher’s hidden layers. CoDIR can be readily applied to compress large-scale language models in both pre-training and finetuning stages, and achieves superb performance on the GLUE benchmark, outperforming state-of-the-art compression methods.
Anthology ID:
2020.emnlp-main.36
Volume:
Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing (EMNLP)
Month:
November
Year:
2020
Address:
Online
Venue:
EMNLP
SIG:
Publisher:
Association for Computational Linguistics
Note:
Pages:
498–508
Language:
URL:
https://aclanthology.org/2020.emnlp-main.36
DOI:
10.18653/v1/2020.emnlp-main.36
Bibkey:
Cite (ACL):
Siqi Sun, Zhe Gan, Yuwei Fang, Yu Cheng, Shuohang Wang, and Jingjing Liu. 2020. Contrastive Distillation on Intermediate Representations for Language Model Compression. In Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing (EMNLP), pages 498–508, Online. Association for Computational Linguistics.
Cite (Informal):
Contrastive Distillation on Intermediate Representations for Language Model Compression (Sun et al., EMNLP 2020)
Copy Citation:
PDF:
https://preview.aclanthology.org/auto-file-uploads/2020.emnlp-main.36.pdf
Video:
 https://slideslive.com/38939327
Code
 intersun/CoDIR
Data
CoLAGLUEMRPCMultiNLIQNLISST