Loading timeline…
20182010s
BERT (Bidirectional Encoder Representations)
A landmark pre-trained open-source transformer model for understanding language context.
Why It Was Important
Unlike early unidirectional models, BERT read sentences in both directions simultaneously to understand the deep context of each word. Released openly by Google, it was rapidly integrated into Google Search's core algorithm and became the starting baseline for thousands of enterprise NLP tasks globally.
Who Invented It
Jacob Devlin et al. (Google)
NLP researchers focused on bidirectional pre-training.
Applications
- Google Search Ranking
- Question Answering
- Sentiment Analysis Baseline
Key Papers
- BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding
Jacob Devlin et al. · NAACL 2019 · 2018
Videos
Transformer models and BERT model: Overview
Google Cloud Tech