Introduction to Transformers: The Architecture Behind BERT and GPT
中文摘要
本文介绍了作为 BERT 和 GPT 基础的 Transformer 架构，并探讨了它如何解决 Word2Vec 等传统 NLP 技术的局限性。
English Summary
This article introduces the Transformer architecture, the foundation of BERT and GPT, explaining how it overcomes the limitations of earlier NLP techniques like Word2Vec.
原文节选
Every technique we’ve covered so far in this series — Bag of Words, TF-IDF, Word2Vec, even sequence labeling models shares one limitation… Continue reading on Medium »