Char-level CNN model for text classification by PyTorch

This repo contains a PyTorch implementation of a char-level CNN model for text classification.

The model architecture comes from this paper:https://arxiv.org/pdf/1509.01626.pdf

Structure of the code

At the root of the project, you will see:

├── pyCharCnn
|  └── callback
|  |  └── lrscheduler.py　　
|  |  └── trainingmonitor.py　
|  |  └── ...
|  └── config
|  |  └── basic_config.py #a configuration file for storing model parameters
|  └── dataset　　　
|  └── io　　　　
|  |  └── dataset.py　　
|  |  └── data_transformer.py　　
|  └── model
|  |  └── nn　
|  └── output #save the ouput of model
|  └── preprocessing #text preprocessing 
|  └── train #used for training a model
|  |  └── trainer.py 
|  |  └── ...
|  └── utils # a set of utility functions
├── train_cnn.py

Dependencies

• csv
• tqdm
• numpy
• pickle
• scikit-learn
• PyTorch 1.0
• matplotlib

How to use the code

1. Download the AG News Topic Classification Dataset from coming and place it into the /pyCharCnn/dataset/raW directory.
2. Modify configuration information in pyCharCnn/config/basic_config.py(the path of data,...).
3. run python train_cnn.py.

Result

coming soon....

training Figure

coming soon