✍️ Next Word Prediction with LSTM

A Streamlit web app that predicts the next word in a given sequence using a trained LSTM model on a text corpus.
Helps demonstrate how Recurrent Neural Networks can learn language patterns and complete sentences.

---

📚 Project Overview

This project allows users to input the beginning of a sentence, and the app predicts the next most likely word based on a trained LSTM model.
It uses Keras Tokenizer for text processing and Streamlit for building a clean, interactive interface.

---

🚀 Features

• Next Word Prediction: Predicts the most probable next word given a sequence.
• Simple UI: Built with Streamlit for easy and intuitive user interaction.
• Language Modeling: Demonstrates basic NLP using deep learning techniques.
• Custom Trained Model: Model trained on a corpus like Shakespeare’s works or any large text dataset.

---

🛠️ Technologies Used

• Python
• TensorFlow / Keras
• Streamlit
• Pickle (for loading tokenizer)

---

🧠 How It Works

• Loads a pre-trained LSTM model (next_word_lstm.h5) and corresponding tokenizer.
• The user enters a sequence of words.
• The sequence is tokenized and padded to match the model's expected input size.
• The model predicts the next word’s token, which is then mapped back to the actual word.
• The result is displayed instantly on the web app.

---

⚙️ How to Run Locally

1. Clone this repository:

   git clone https://github.com/your-username/next-word-prediction-lstm.git
   cd next-word-prediction-lstm

2. Install the required packages:

   pip install tensorflow streamlit numpy

3. Ensure you have the necessary files:

   • next_word_lstm.h5 (the trained model)
   • tokenizer.pickle (the tokenizer used during model training)

4. Run the Streamlit app:

   streamlit run app.py

5. Open the app in your browser at:

   http://localhost:8501
   

---

📂 Project Structure

├── app.py                  # Main Streamlit application
├── next_word_lstm.h5        # Pre-trained LSTM model
├── tokenizer.pickle         # Tokenizer used for text encoding
└── README.md                # Project documentation

---

📢 Future Improvements

• Predict multiple next words (not just one word).
• Fine-tune on a larger modern dataset (e.g., Wikipedia, Reddit).
• Improve UI with multiple prediction options.
• Deploy online with Streamlit Cloud or Hugging Face Spaces.

---

🤝 Contributing

Pull requests are welcome!
If you find any bugs or have suggestions for improvement, feel free to open an issue or submit a PR.

---

📜 License

This project is open source under the MIT License.

---

🌟 Show Your Support

If you like this project, don't forget to ⭐ star the repository!

---