Fine-Tuning DistilBERT for Sentiment Analysis

This repository contains the code and workflow for fine-tuning the DistilBERT (distilbert-base-uncased) model from Hugging Face on a sentiment analysis task. The dataset used for training is sourced from Kaggle.

🚀 Features

Fine-tuning distilbert-base-uncased for sentiment classification. Data preprocessing and tokenization using Hugging Face Transformers. Model training and evaluation. Inference script to predict sentiment on new text samples.

📂 Dataset

The 3 datasets used for fine-tuning is available on Kaggle. You can download it using below links:

1. IMDB dataset (Sentiment analysis) in CSV format link
2. Sentiment Analysis Dataset link
3. Stock News Sentiment Analysis(Massive Dataset) link
4. final dataset in Huggingface link

📊 Data Preprocessing & Visualization

The dataset is cleaned, preprocessed, and visualized using Pandas, Matplotlib, and Seaborn. Open and run the notebook:

📜 Notebook: notebooks/data_preprocessing.ipynb

📦 Installation

Clone the repository and install the required dependencies:

``` python 
git clone https://github.com/KaushiML3/Fine-tuning-a-LLM-for-sentiment-analysis.git
cd your-repo-name
pip install -r requirements.txt
```

🛠 Training the Model

The DistilBERT model is fine-tuned using Hugging Face's Transformers library. Training includes learning rate scheduling, and evaluation metrics. Open and run the notebook:

1. 📜 Notebook: notebook/Fine tune LLM with LoRA for sentiment analysis.ipynb

2.Alternatively, run the training script:

``` python 
python train.py
```

🔍 Inference

To test the model on new text inputs, run:

``` python 
    python app.py 
```

• sentiment analysis DistilBERT model demo

📄 Acknowledgments

1. Hugging Face for the DistilBERT model.
2. Kaggle for the dataset.