This repository provides implementations of various deep learning models for Chinese text classification including TextCNN, TextRNN, FastText, TextRCNN, BiLSTM with Attention, DPCNN, and Transformer. It uses character-level and word-level input with pre-trained Chinese word embeddings from Sogou news corpus. The library includes benchmark results on THUCNews dataset across 10 categories, achieving up to 94.83% accuracy with BERT.