Text Classification Models

This repository is forked from https://github.com/TobiasLee/Text-Classification. It adds a commandline interface for convenient model training, evaluation and prediction and updates the code to be compatible with tensorflow 2. The base_model.py and base_train.py modules have been adapted from https://github.com/MrGemy95/Tensorflow-Project-Template

Dependencies

• python >= 3.7
• tensorflow >= 2.0
• scikit-learn
• click
• pandas
• bunch
• tqdm

Very Short Rundown

The models work with csv files containing the columns "title", "content" and "class". Training a model works like this:

python -m cli [modelname] train [training.csv] --model-dir path/to/model/dir

For evaluation after training a model and specifying "path/to/model/dir" as model directory:

python -m cli [modelname] eval [eval.csv] --model-dir path/to/model/dir

For prediction after training a model and specifying "path/to/model/dir" as model directory:

python -m cli [modelname] predict [predict.csv] --model-dir path/to/model/dir