BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding

TL;DR

BERT introduces a deep bidirectional transformer-based model pre-trained on unlabeled text, achieving state-of-the-art results across multiple NLP tasks with minimal task-specific modifications.

Contribution

It presents a novel pre-training method for deep bidirectional representations that significantly improves performance on various NLP benchmarks.

Findings

Achieves new state-of-the-art results on eleven NLP tasks.

Pushes GLUE score to 80.5%.

Improves SQuAD v1.1 and v2.0 F1 scores substantially.

Abstract

We introduce a new language representation model called BERT, which stands for Bidirectional Encoder Representations from Transformers. Unlike recent language representation models, BERT is designed to pre-train deep bidirectional representations from unlabeled text by jointly conditioning on both left and right context in all layers. As a result, the pre-trained BERT model can be fine-tuned with just one additional output layer to create state-of-the-art models for a wide range of tasks, such as question answering and language inference, without substantial task-specific architecture modifications. BERT is conceptually simple and empirically powerful. It obtains new state-of-the-art results on eleven natural language processing tasks, including pushing the GLUE score to 80.5% (7.7% point absolute improvement), MultiNLI accuracy to 86.7% (4.6% absolute improvement), SQuAD v1.1 question answering Test F1 to 93.2 (1.5 point absolute improvement) and SQuAD v2.0 Test F1 to 83.1 (5.1 point absolute improvement).