Keras-TextClassification

Why we included this project

Anyone building Chinese text classifiers and wanting to compare architectures side by side will find a handy reference here. The Keras library bundles FastText, TextCNN, TextRNN, RCNN, DPCNN, VDCNN, HAN, capsule networks, and Transformer encoders, plus fine-tuning examples for BERT, XLNet, and ALBERT. Because most models build on shared embedding and network-layer base classes, they run with the same setup, so swapping one for another does not mean restructuring your data pipeline. Working examples for multi-label classification and sentence-pair similarity cover needs beyond plain single-label sorting. Treat it as a study and experimentation toolkit rather than a maintained production framework: some pretrained embeddings and datasets must be downloaded separately, and a few models are still marked as work in progress.