Open PDF if the ipynb file doesn't show up. The ipynb works when you download and open locally.

BERT Sentiment Analysis using Hugging Face

This repository contains an end-to-end implementation of sentiment analysis using a pre-trained BERT model fine-tuned with the Hugging Face Transformers library. The project demonstrates a complete NLP workflow, starting from exploratory data analysis to model training, evaluation, and saving the trained model for reuse.

---

Project Overview

The goal of this project is to build a robust sentiment classification model using BERT. The notebook follows best practices for modern NLP pipelines, including stratified data splitting, Hugging Face Dataset integration, metric-based evaluation, and efficient model fine-tuning.

---

Dataset

• The dataset is sourced from GitHub
• It contains labeled text data for sentiment classification
• The data is split into train, validation, and test sets using stratified sampling to preserve class distribution

---

Workflow

The project follows the steps below:

1. Library Setup

   • PyTorch
   • Hugging Face Transformers & Datasets
   • Scikit-learn, evaluate
   • NumPy, Pandas, matplotlib, seaborn

2. Exploratory Data Analysis (EDA)

   • Basic inspection of text samples
   • Label distribution analysis

3. Stratified Train–Validation–Test Split

   • Ensures balanced class representation across splits

4. Hugging Face Dataset Conversion

   • Conversion from Pandas DataFrame to Dataset and DatasetDict

5. Label Encoding

   • Creation of label2id and id2label mappings for model compatibility

6. Model Selection

   • Pre-trained BERT model for sequence classification
   • Loaded using Hugging Face Transformers

7. Tokenization

   • Tokenization using BERT tokenizer
   • Truncation and padding applied
   • Removal of unnecessary columns to reduce memory usage

8. Training Configuration

   • TrainingArguments configured for:
     • Evaluation during training
     • Logging
     • Model checkpointing

9. Evaluation Metrics

   • Accuracy
   • F1-score
   • Precision

10. Model Training

    • Fine-tuning performed using Hugging Face Trainer API

11. Prediction and Evaluation

    • Model evaluated on the test dataset
    • Metrics reported using standard NLP evaluation practices

12. Model Saving

    • Fine-tuned model and tokenizer saved for future inference or deployment

---

Results

The fine-tuned BERT model achieved the following performance on the test set:

• Accuracy: 93.18%
• F1 Score: 93.19%
• Test Loss: 0.1966
• Evaluation Runtime: 15.95 seconds
• Samples per Second: 200.59

These results demonstrate strong generalization and effective fine-tuning of the BERT model for sentiment classification.

---

Technologies Used

• Python
• PyTorch
• Hugging Face Transformers
• Hugging Face Datasets
• Scikit-learn
• Google Colab (GPU)

---