Skip to content

Latest commit

 

History

History
30 lines (17 loc) · 884 Bytes

README.md

File metadata and controls

30 lines (17 loc) · 884 Bytes

Word2vec_With_Gensim

What's Word2Vec

it's a 2 layer neural network to generate word embeddings given a text corpus . and here we talk about the famous exemple :

King - Man + Women = Queen

we can conclude that 2 words have the same Corpus , have similar embedding .

Why we use Word2vec :

  • the machine dont understand text , so we should convert the text into numbers , thats why we should convert text into numeric input .

  • The second goal is to convert high dimensional feature to low dimentional feature.

  • Preserves relationship between words

  • better results in lots of deep learning applications

Model Architecture for Word2Vec :

  • CBOW (Continuous bag of words) : Predict The target word from the context

  • Skip Gram : Predict the context from the Traget Word

Choice Of Model :

  • CBOW : Small Corpus / Fast
  • Skip Gram : Large corpus / Slow