✳flâneur — a map of the web's best reading
AutoEncoder (三)- Self Attention、Transformer | by Leyan Bin Veon | NLP-ML筆記 | Medium
medium.com · 673 words · saved by 1 readers
Part 1中,我們學到attention model是如何運作的,缺點就是不能平行化,且忽略了輸入句中文字間和目標句中文字間的關係。
AutoEncoder (三)- Self Attention、Transformer 前言 Moris 16 min read · Jun 24, 2019 -- Share Part 1 中,我們學到attention model是如何運作的,缺點就是不能平行化,且忽略了輸入句中文字間和目標句中文字間的關係。 為了解決此問題,2017年,Self attention誕生了。 Press enter or click to view image in full size 圖. 1. Attention model 四格漫畫 Self Attention Self attention是Google在 “ Attention is all you need ”論文中提出的”The transformer”模型中主要的概念之一。 如下圖所示,encoder讀入輸入數據,利用層層疊加的Self-Attention機制對每一個詞得到新的考慮了上下文信息的表徵。Decoder也利用類似的Self-Attention機制,但它不僅僅看之前產生的輸出的文字,而且還要attend encoder的輸出。以上步驟如下動圖所示: 了解了模型大致原理,我們可以詳細的看一下究竟Self-Attention結構是怎樣的。其基本結構如下 Press enter or click to view image i
Explore this link on the map →saved by
related reading
- The Illustrated Transformer – Jay Alammar – Visualizing machine learning one concept at a time.jalammar.github.io
- The Illustrated Transformer – Jay Alammar – Visualizing machine learning one concept at a time.jalammar.github.io
- Transformers Explained Visually (Part 3): Multi-head Attention, deep dive | Towards Data Sciencetowardsdatascience.com
- transformer_attention.pdfarxiv.org
- Everything About Transformerskrupadave.com
- Transformer (deep learning) - Wikipediaen.wikipedia.org
- 11.7. The Transformer Architecture — Dive into Deep Learning 1.0.3 documentationd2l.ai
- A Mathematical Framework for Transformer Circuitstransformer-circuits.pub
- Attention is all you need: Discovering the Transformer paper | Towards Data Sciencetowardsdatascience.com
- 1706.03762arxiv.org
- Transformers from scratch | peterbloem.nlpeterbloem.nl
- Some Intuition on Attention and the Transformereugeneyan.com