nlp-notebook

Why we included this project

Developers and students working through Chinese NLP in PyTorch will find this notebook a practical companion. It walks through the full pipeline: new-word discovery, word2vec and GloVe embeddings, text classification from TextCNN to distilled BERT, several BERT-based NER setups, abstractive and extractive summarization, sentence similarity, and relation extraction. Each section is self-contained with its own data and training scripts, and the classification models use Optuna for hyperparameter tuning, so you can see how real tuning fits in. It is less a drop-in production library and more a study resource that shows how each technique is actually implemented, which makes it useful for building intuition before you design your own solution.