Next Word Predictor using BiLSTM (PyTorch)

📌 Overview

This repository contains the implementation of a Next Word Prediction model using Bidirectional Long Short-Term Memory (BiLSTM) in PyTorch. The model is trained on a dataset of Medium article titles to predict the next word in a given sentence.

📂 Dataset

The dataset used is the Medium Articles Dataset from Kaggle, containing article titles. The data is preprocessed using NLTK tokenization.

🔧 Installation & Dependencies

To set up the project locally, follow these steps:

1️⃣ Clone the Repository

git clone https://github.com/yourusername/Next-word-predictor-BiLSTM-pytorch.git
cd Next-word-predictor-BiLSTM-pytorch

📊 Model Architecture

• Embedding Layer: Converts words into dense vector representations.
• BiLSTM Layer: Captures long-term dependencies in both forward and backward directions.
• Fully Connected Layer: Outputs the probability distribution of the next word.

📈 Results

The model achieves reasonable accuracy on predicting the next word given an input phrase. More evaluation metrics and fine-tuning results will be added.

👤 Author

Anshul Katiyar
GitHub: Anshul21107
LinkedIn: Anshul Katiyar

---