RESEARCH

NLP

MixKD: Towards Efficient Distillation of Large-scale Language Models

March 17, 2021

Abstract

Large-scale language models have recently demonstrated impressive empirical performance. Nevertheless, the improved results are attained at the price of bigger models, more power consumption, and slower inference, which hinder their applicability to low-resource (both memory and computation) platforms. Knowledge distillation (KD) has been demonstrated as an effective framework for compressing such big models. However, large-scale neural network systems are prone to memorize training instances, and thus tend to make inconsistent predictions when the data distribution is altered slightly. Moreover, the student model has few opportunities to request useful information from the teacher model when there is limited task-specific data available. To address these issues, we propose MixKD, a data-agnostic distillation framework that leverages mixup, a simple yet efficient data augmentation approach, to endow the resulting model with stronger generalization ability. Concretely, in addition to the original training examples, the student model is encouraged to mimic the teacher's behavior on the linear interpolation of example pairs as well. We prove from a theoretical perspective that under reasonable conditions MixKD gives rise to a smaller gap between the generalization error and the empirical error. To verify its effectiveness, we conduct experiments on the GLUE benchmark, where MixKD consistently leads to significant gains over the standard KD training, and outperforms several competitive baselines. Experiments under a limited-data setting and ablation studies further demonstrate the advantages of the proposed approach.

Download the Paper

AUTHORS

Written by

Kevin J Liang

Weituo Hao

Dinghan Shen

Yufan Zhou

Weizhu Chen

Changyou Chen

Lawrence Carin

Publisher

ICLR 2021

Related Publications

September 10, 2019

NLP

Bridging the Gap Between Relevance Matching and Semantic Matching for Short Text Similarity Modeling | Facebook AI Research

A core problem of information retrieval (IR) is relevance matching, which is to rank documents by relevance to a user’s query. On the other hand, many NLP problems, such as question answering and paraphrase identification, can be considered…

Jinfeng Rao, Linqing Liu, Yi Tay, Wei Yang, Peng Shi, Jimmy Lin

September 10, 2019

June 11, 2019

NLP

COMPUTER VISION

Adversarial Inference for Multi-Sentence Video Description | Facebook AI Research

While significant progress has been made in the image captioning task, video description is still in its infancy due to the complex nature of video data. Generating multi-sentence descriptions for long videos is even more challenging. Among the…

Jae Sung Park, Marcus Rohrbach, Trevor Darrell, Anna Rohrbach

June 11, 2019

May 17, 2019

NLP

Unsupervised Hyper-alignment for Multilingual Word Embeddings | Facebook AI Research

We consider the problem of aligning continuous word representations, learned in multiple languages, to a common space. It was recently shown that, in the case of two languages, it is possible to learn such a mapping without supervision. This…

Jean Alaux, Edouard Grave, Marco Cuturi, Armand Joulin

May 17, 2019

July 27, 2019

NLP

Unsupervised Question Answering by Cloze Translation | Facebook AI Research

Obtaining training data for Question Answering (QA) is time-consuming and resource-intensive, and existing QA datasets are only available for limited domains and languages. In this work, we explore to what extent high quality training data is…

Patrick Lewis, Ludovic Denoyer, Sebastian Riedel

July 27, 2019

Help Us Pioneer The Future of AI

We share our open source frameworks, tools, libraries, and models for everything from research exploration to large-scale production deployment.