hansw90/nlp

자연어처리 학습 및 기본적인 NN 학습 repository

자연어처리 학습 및 논문 정리를 위한 페이지입니다.

---

NLP 폴더

• Bert_EmbeddingLayer.ipynb : 벌트 개념정리

  • Token Embedding
  • Segment Embeddings (부분 임베딩?)
  • Positional Embedding
  • Positional Embedding
    https://github.com/hansw90/NLP-natural-language-processing-/blob/master/NLP/Bert_EmbeddingLayer.ipynb

• BertChat_code.ipynb : Bert를 이용한 chat 프로그래밍 코드 입니다. 현재 코드 분석중
  https://github.com/hansw90/NLP-natural-language-processing-/blob/master/NLP/BertChat_code.ipynb

• Bert(Kaggle).ipynb : kaggle을 통해 Bert를 공부중입니다.
  https://github.com/hansw90/NLP-natural-language-processing-/blob/master/NLP/Bert(Kaggle).ipynb

• KoNLPy 개념및 간단 코드 정리
  https://github.com/hansw90/NLP-natural-language-processing-/blob/master/NLP/Konlpy.ipynb

• Lstm 코드로 정리
  https://github.com/hansw90/NLP-natural-language-processing-/blob/master/NLP/LSTM.py

• RNN01.ipynb : RNN 개념정리

  • RNN 이란?
  • RNN 으로 무엇을 할 수 있을까?
  • RNN 학습하기
  • RNN 의 확장된 모델들
    https://github.com/hansw90/NLP-natural-language-processing-/blob/master/NLP/RNN01.ipynb

• RNN02_Vanishing_Gradient.ipynb : RNN 의 단점 기울기 손실에 대해 알아보자.

  • BPTT (backpropagation through time
  • Vanishing Gradient Problem
    https://github.com/hansw90/NLP-natural-language-processing-/blob/master/NLP/RNN02_Vanishing_Gradient.ipynb

• SimpleRNN.ipynb : NN, RNN 구현 사실 신경망이 아닌 그냥 하드코딩, 그냥 RNN은 이전 정보를 가지고 새로운 정보를 도출해낸다는 느낌을 주기 위해 구현

  • SimpleNN
  • SimpleRNN
    https://github.com/hansw90/NLP-natural-language-processing-/blob/master/NLP/SimpleRNN.ipynb

• VanilaRNN
  https://github.com/hansw90/NLP-natural-language-processing-/blob/master/NLP/VanilaRNN.ipynb

• Word_embedding_basic.ipynb

  • 쉽게 쓰인 word2vec
  • 단어 임베딩(Word Embedding) 맛보기
  • Sparse vs. Dense Representataions
  • Sparse representataion
  • Dense representation
  • Dense representation (distributed representaion의 장점)
  • Word2Vec
  • CBOW

https://github.com/hansw90/NLP-natural-language-processing-/blob/master/NLP/Word_embedding_basic.ipynb

• WordEmbedding

  • 영어 Word2vec 만들기
  • 글로브(GloVe)
  • Using pre-trained word embeddings in a Keras model
  • ELMO
    https://github.com/hansw90/NLP-natural-language-processing-/blob/master/NLP/WordEmbedding.ipynb

• fastText (FastText, Word representation using subword)

  • Introduction
  • Out of vocabuary, infrequent words (Word2Vec)
  • Subword representation
  • negative samples
  • 한국어를 위한 FastText: 초/중/종성 분리 (code)
  • Package (fastText)
    https://github.com/hansw90/NLP-natural-language-processing-/blob/master/NLP/WordEmbedding_fastText.ipynb

• Word Piece Model

  • Word piece, unit of words
  • Word Piece Model (sentencepiece) tokenizer
  • Byte-pair Encoding (BPE)
    https://github.com/hansw90/NLP-natural-language-processing-/blob/master/NLP/Word_Piece_Embedding.ipynb

• Document Similarity : 문서 유사도

  • 유클리드 거리
  • 자카드 유사도

https://github.com/hansw90/NLP/blob/master/NLP/DocumentSimilarity.ipynb

• BERT-BTC

  • The A-to-Z guide on how you can use Google’s BERT for binary text classification tasks. I’ll be aiming to explain, as simply and straightforwardly as possible, how to fine-tune a BERT model (with PyTorch) and use it for a binary text classification task.

• Attention 완전 뽀개기 https://github.com/hansw90/NLP/blob/master/NLP/Attention_PaperDetail.ipynb

• A Structured Self-Attentive Sentence Embedding