Skip to content

Latest commit

 

History

History
52 lines (22 loc) · 725 Bytes

File metadata and controls

52 lines (22 loc) · 725 Bytes

TextCNN

A picture is worth a thousand words !

Dataset

Sentences are subset of **THUCTC **(THU Chinese Text Classification), http://thuctc.thunlp.org/, format as fastText.

Train

python train.py -h

Evaluate

python eval.py -h

Predict

python predict.py -h

Your text must be separated by space and longer than the max kernel size (Reference config.py).

Requirement

License

TextCNN is released under the [Apache 2.0 license]

Chinese Document

通用文本分类器.