Sentimental Analysis Using Bert Transformer model

Project workflow
├── config.py (It contain all parameter of Bert model and path for dataset) 
│    ├── IMDB Dataset.csv (It is Amazon review dataset contain review and sentiment associate with it)
│    ├── dataset.py (Load dataset, preprocessing and input for model)
│    ├── model.py (It load pretained model over dataset)
│    ├── engine.py (It contain Bert model)
│    ├── train.py (It load train, eval function for training and run it for training model)
│    └── Train & Predict.ipynb (It run all the python scripts for training and prediction in Jupyter Notebook)

Aim

Build End-to-End Machine learning pipline for preprocessing, exploratory data analysis, modelling, deployment.

Dataset

Train over Amazon review dataset using Bert Transformer model and use f1-score for evalution of model. Dataset

Save model

Trained model can be download Model

Tools:

Flask, Transformers, pytorch, HTML, CSS, Javascripts, AWS EC2, nltk

Prediction

Deployment code is available in deploy branch containing web application integrated with Flask app Deploy

Result