<!-- WEHUB_ZH_README -->
> [!NOTE]
> 本文档由 WeHub 基于上游 README 翻译整理，属于社区翻译，非官方中文文档。  
> [English](./README.en.md) · [原始项目](https://github.com/graykode/nlp-tutorial) · [上游 README](https://github.com/graykode/nlp-tutorial/blob/HEAD/README.md)  
> 原作者、版权与许可证归属以原始项目及本仓库 LICENSE 文件为准。

## nlp-tutorial

<p align="center"><img width="100" src="https://upload.wikimedia.org/wikipedia/commons/thumb/1/11/TensorFlowLogo.svg/225px-TensorFlowLogo.svg.png" />  <img width="100" src="https://media-thumbs.golden.com/OLqzmrmwAzY1P7Sl29k2T9WjJdM=/200x200/smart/golden-storage-production.s3.amazonaws.com/topic_images/e08914afa10a4179893eeb07cb5e4713.png" /></p>

`nlp-tutorial` 是一份面向学习 NLP（自然语言处理，Natural Language Processing）者的教程，使用 **Pytorch** 实现。NLP 中的大多数模型均用不到 **100 行**代码实现。（注释和空行除外）

- [08-14-2020] 旧版 TensorFlow v1 代码已归档至 [archive 文件夹](archive)。为便于初学者阅读，仅支持 PyTorch 1.0 及以上版本。


## 课程大纲 -（示例用途）

#### 1. 基础嵌入模型（Embedding Model）

- 1-1. [NNLM（神经网络语言模型，Neural Network Language Model）](1-1.NNLM) - **预测下一个词（Predict Next Word）**
  - 论文 -  [A Neural Probabilistic Language Model(2003)](http://www.jmlr.org/papers/volume3/bengio03a/bengio03a.pdf)
  - Colab - [NNLM.ipynb](https://colab.research.google.com/github/graykode/nlp-tutorial/blob/master/1-1.NNLM/NNLM.ipynb)
- 1-2. [Word2Vec（Skip-gram）](1-2.Word2Vec) - **词嵌入与图形展示（Embedding Words and Show Graph）**
  - 论文 - [Distributed Representations of Words and Phrases
    and their Compositionality(2013)](https://papers.nips.cc/paper/5021-distributed-representations-of-words-and-phrases-and-their-compositionality.pdf)
  - Colab - [Word2Vec.ipynb](https://colab.research.google.com/github/graykode/nlp-tutorial/blob/master/1-2.Word2Vec/Word2Vec_Skipgram(Softmax).ipynb)
- 1-3. [FastText（应用层，Application Level）](1-3.FastText) - **句子分类（Sentence Classification）**
  - 论文 - [Bag of Tricks for Efficient Text Classification(2016)](https://arxiv.org/pdf/1607.01759.pdf)
  - Colab - [FastText.ipynb](https://colab.research.google.com/github/graykode/nlp-tutorial/blob/master/1-3.FastText/FastText.ipynb)


#### 2. CNN（卷积神经网络，Convolutional Neural Network）

- 2-1. [TextCNN](2-1.TextCNN) - **二分类情感分析（Binary Sentiment Classification）**
  - 论文 - [Convolutional Neural Networks for Sentence Classification(2014)](http://www.aclweb.org/anthology/D14-1181)
  - [TextCNN.ipynb](https://colab.research.google.com/github/graykode/nlp-tutorial/blob/master/2-1.TextCNN/TextCNN.ipynb)


#### 3. RNN（循环神经网络，Recurrent Neural Network）

- 3-1. [TextRNN](3-1.TextRNN) - **预测下一步（Predict Next Step）**
  - 论文 - [Finding Structure in Time(1990)](http://psych.colorado.edu/~kimlab/Elman1990.pdf)
  - Colab - [TextRNN.ipynb](https://colab.research.google.com/github/graykode/nlp-tutorial/blob/master/3-1.TextRNN/TextRNN.ipynb)
- 3-2. [TextLSTM](https://github.com/graykode/nlp-tutorial/tree/master/3-2.TextLSTM) - **自动补全（Autocomplete）**
  - 论文 - [LONG SHORT-TERM MEMORY(1997)](https://www.bioinf.jku.at/publications/older/2604.pdf)
  - Colab - [TextLSTM.ipynb](https://colab.research.google.com/github/graykode/nlp-tutorial/blob/master/3-2.TextLSTM/TextLSTM.ipynb)
- 3-3. [Bi-LSTM](3-3.Bi-LSTM) - **在长句中预测下一个词（Predict Next Word in Long Sentence）**
  - Colab - [Bi_LSTM.ipynb](https://colab.research.google.com/github/graykode/nlp-tutorial/blob/master/3-3.Bi-LSTM/Bi_LSTM.ipynb)


#### 4. 注意力机制（Attention Mechanism）

- 4-1. [Seq2Seq](4-1.Seq2Seq) - **词语替换（Change Word）**
  - 论文 - [Learning Phrase Representations using RNN Encoder–Decoder
    for Statistical Machine Translation(2014)](https://arxiv.org/pdf/1406.1078.pdf)
  - Colab - [Seq2Seq.ipynb](https://colab.research.google.com/github/graykode/nlp-tutorial/blob/master/4-1.Seq2Seq/Seq2Seq.ipynb)
- 4-2. [Seq2Seq with Attention](4-2.Seq2Seq(Attention)) - **机器翻译（Translate）**
  - 论文 - [Neural Machine Translation by Jointly Learning to Align and Translate(2014)](https://arxiv.org/abs/1409.0473)
  - Colab - [Seq2Seq(Attention).ipynb](https://colab.research.google.com/github/graykode/nlp-tutorial/blob/master/4-2.Seq2Seq(Attention)/Seq2Seq(Attention).ipynb)
- 4-3. [Bi-LSTM with Attention](4-3.Bi-LSTM(Attention)) - **二分类情感分析（Binary Sentiment Classification）**
  - Colab - [Bi_LSTM(Attention).ipynb](https://colab.research.google.com/github/graykode/nlp-tutorial/blob/master/4-3.Bi-LSTM(Attention)/Bi_LSTM(Attention).ipynb)


#### 5. 基于 Transformer 的模型

- 5-1.  [The Transformer](5-1.Transformer) - **机器翻译（Translate）**
  - 论文 - [Attention Is All You Need(2017)](https://arxiv.org/abs/1706.03762)
  - Colab - [Transformer.ipynb](https://colab.research.google.com/github/graykode/nlp-tutorial/blob/master/5-1.Transformer/Transformer.ipynb), [Transformer(Greedy_decoder).ipynb](https://colab.research.google.com/github/graykode/nlp-tutorial/blob/master/5-1.Transformer/Transformer(Greedy_decoder).ipynb)
- 5-2. [BERT](5-2.BERT) - **下一句分类与掩码词预测（Classification Next Sentence & Predict Masked Tokens）**
  - 论文 - [BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding(2018)](https://arxiv.org/abs/1810.04805)
  - Colab - [BERT.ipynb](https://colab.research.google.com/github/graykode/nlp-tutorial/blob/master/5-2.BERT/BERT.ipynb)


## 依赖项

- Python 3.5+
- Pytorch 1.0.0+


## 作者

- Tae Hwan Jung(Jeff Jung) @graykode
- 作者邮箱：nlkey2022@gmail.com
- 感谢 [mojitok](http://mojitok.com/) 担任 NLP 研究实习生。