Amazon Sentiment Analysis-Before EDA.ipynb contains the code for the training of the model before using Easy Data Augmentation. It has an accuracy of 60%.
Amazon Sentiment Analysis-After EDA (Best Version).ipynb contains the code for the training of the model after using Easy Data Augmentation. It boosts the accuracy to 78.7%. The code is also commented unlike the other notebook, so use this for further reference.
Took around 48 hrs to train both models. I used P4000 GPU with 8 CPUs and 30 GB of RAM from Paperspace's Gradient ML Platform.