Repository containing the code to develop a Neural based text classifier.
In the repository there are various models implemented for text classification. In order to access a ready-to-explore version one can have a look at the notebooks provided. Models are quite heavy and memory consuming, so it is really advised to use a GPU machine to run their training tasks.
| Model | Demo | Details | CLI | Accuracy score on AG news dataset |
|---|---|---|---|---|
CNN TextClassifier |
 |
Classify texts with labels from the AG news database making use of a convolutional neural network. | python3 -m train -c config.yaml |
90.71 |
| webApp | ||||
| source | ||||
BERT TextClassifier |
|
Classify texts with labels from the AG news database making use of an attention model, based on BERT. | python3 -m train -c config.yaml |
93.95 |
| webApp | ||||
| source |