This repository contains the code for sentiment analysis on Amazon product reviews using a Transformer-based model.
The model predicts whether a product review is positive or negative with probability scores and visual feedback.
The model was trained using a BERT architecture fine-tuned on Amazon review data for sentiment classification.
The actual trained model files are hosted on Hugging Face to avoid large file uploads on GitHub.
- Open the Kaggle notebook
amazon-product-review-nlp.ipynb - The notebook loads the model files from Hugging Face and performs inference on input reviews.
The following model files are hosted on Hugging Face:
The notebook downloads these files directly for inference.
- This repo contains only the code and documentation.
- The heavy model files are loaded from Hugging Face at runtime.
- For a live demo or deployment, you can use Kaggle notebooks as the platform.
Created by Arpita Rout