The package currently uses Multi Layered Perceptron or ANN as the highest level model for the classification tasks. Several other advanced model architectures like RNN, GRU, LSTM and Transformers have been proven to work better for the concerned NLP and ML tasks and thus can be implemented for the same.
This would definitely help better the results across sentiment analysis, tagging and classification tasks. I am ready to contribute by implementing and integrating this task, if the authors are interested.
The package currently uses Multi Layered Perceptron or ANN as the highest level model for the classification tasks. Several other advanced model architectures like RNN, GRU, LSTM and Transformers have been proven to work better for the concerned NLP and ML tasks and thus can be implemented for the same.
This would definitely help better the results across sentiment analysis, tagging and classification tasks. I am ready to contribute by implementing and integrating this task, if the authors are interested.