Sentiment Analysis on Product Reviews using Hugging Face Transformers

This project fine-tunes DistilBERT to perform binary sentiment classification (positive/negative) on the IMDb movie reviews dataset using the Hugging Face transformers and datasets libraries.

---

📌 Features

• Fine-tunes a pre-trained transformer model on real-world text data.
• Uses Hugging Face's Trainer API for training and evaluation.
• Includes accuracy and F1-score metrics.
• Inference pipeline for testing custom reviews.
• Fully reproducible in Google Colab.

---

📂 Example Inference

from transformers import pipeline

pipeline = pipeline("text-classification", model="mhemon/seintiment-distilbert-base-uncased", device="cuda")

pipeline("The product quality is amazing and delivery was fast!")
# [{'label': 'POSITIVE', 'score': 0.7757363319396973}]

📊 Result

Metric	Score
Accuracy	0.85
F1 Score	0.85

Colab File

https://colab.research.google.com/drive/1hycSErDqoGt1mWz20Al3fRX6EbDbW2nF?usp=sharing