MobileBERT: a Compact Task-Agnostic BERT for Resource-Limited Devices

Code

• github.com/tensorflow/models/tree/master/official/nlp/projects/mobilebert
  Officialtf★ 0
• github.com/tchebonenko/Automated-Topic_Modeling-and-NER
  tf★ 4
• github.com/Milan-Chicago/GLG-Automated-Meta-data-Tagging
  tf★ 2
• github.com/pwc-1/Paper-5/tree/main/mobilebert
  mindspore★ 0
• github.com/nosaydomore/MobileBert_paddle
  paddle★ 0
• github.com/2023-MindSpore-1/ms-code-159
  mindspore★ 0
• github.com/MS-P3/code5/tree/main/mobilebert
  mindspore★ 0

Abstract

Natural Language Processing (NLP) has recently achieved great success by using huge pre-trained models with hundreds of millions of parameters. However, these models suffer from heavy model sizes and high latency such that they cannot be deployed to resource-limited mobile devices. In this paper, we propose MobileBERT for compressing and accelerating the popular BERT model. Like the original BERT, MobileBERT is task-agnostic, that is, it can be generically applied to various downstream NLP tasks via simple fine-tuning. Basically, MobileBERT is a thin version of BERT_LARGE, while equipped with bottleneck structures and a carefully designed balance between self-attentions and feed-forward networks. To train MobileBERT, we first train a specially designed teacher model, an inverted-bottleneck incorporated BERT_LARGE model. Then, we conduct knowledge transfer from this teacher to MobileBERT. Empirical studies show that MobileBERT is 4.3x smaller and 5.5x faster than BERT_BASE while achieving competitive results on well-known benchmarks. On the natural language inference tasks of GLUE, MobileBERT achieves a GLUEscore o 77.7 (0.6 lower than BERT_BASE), and 62 ms latency on a Pixel 4 phone. On the SQuAD v1.1/v2.0 question answering task, MobileBERT achieves a dev F1 score of 90.0/79.2 (1.5/2.1 higher than BERT_BASE).

Tasks

Benchmark Results

Reproductions