I have tried to explore the TorchText library using kaggle dataset https://www.kaggle.com/datasets/jp797498e/twitter-entity-sentiment-analysis.
I only focussed here to explore the different things, not on accuracy of the model. For that, I did not used full dataset and not performed the raw text processing here.
Main code is written in main.py file and data processing code is written in data_process.py file. Main experiment was done in Jupyter notebook rough_experiment.ipynb file, and after that all codes written in organized manner in VS code.
I have made a proper Sentiment Analysis Model using PyTorch with modular coding approach, link is given here https://github.com/Kousik-Sasmal/tweeter-sentiment-using-pytorch-torchtext.