bramdekker/TextCNN

Convolutional Neural Networks for Sentence Classification(https://arxiv.org/abs/1408.5882) .

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

• python 3.6.2

• pytorch 1.0.0

• torchtext 0.4.0

• https://arxiv.org/abs/1408.5882)

License

TextCNN is released under the [Apache 2.0 license]

Chinese Document

通用文本分类器.