🧠 AI Next Word Prediction

╔════════════════════════════════════════════════════════════╗
║  Project     : AI Next Word Prediction                     ║
║  Model       : LSTM RNN Language Model                     ║
║  Dataset     : Shakespeare Hamlet Text                     ║
║  Framework   : TensorFlow / Keras + Streamlit              ║
║  Author      : ASHITOSH                                    ║
╚════════════════════════════════════════════════════════════╝

---

📌 Project Overview

AI Next Word Prediction is a deep learning NLP project that predicts the next word for a given text sequence. The model is trained on Shakespeare's Hamlet text using an LSTM-based Recurrent Neural Network.

The project includes:

• 🧠 A trained LSTM next-word prediction model
• 🔤 A saved tokenizer for text-to-sequence conversion
• 📚 Hamlet text dataset
• 📓 Training and experimentation notebook
• 🚀 A deployable Streamlit web application

---

✨ Features

• ✅ Predicts the next likely word from user input
• ✅ Uses n-gram sequence generation for language modeling
• ✅ Pads sequences before model inference
• ✅ Uses a trained Keras .h5 model
• ✅ Provides a simple Streamlit UI
• ✅ Ready for local running and Streamlit Cloud deployment

---

🗂️ Project Directory

LSTM_RNN_imple/
├── .streamlit/
│   └── config.toml
├── app.py
├── experiments.ipynb
├── hamlet.txt
├── next_word_prediction_model.h5
├── README.md
├── requirements.txt
├── runtime.txt
└── tokenizer.pickle

---

🛠️ Tech Stack

Badge	Library	Purpose
	Python	Core language
	TensorFlow	Deep learning framework
	Keras	Model building API
	Streamlit	Web app interface
	NumPy	Numerical operations
	Pandas	Data manipulation
	Matplotlib	Visualization
	Scikit-learn	Utilities
	NLTK	Text / corpus tools

---

🧪 Model Information

The model pipeline follows these steps:

Step	Description
1	Collect Hamlet text data using NLTK / Gutenberg
2	Tokenize text with tensorflow.keras.preprocessing.text.Tokenizer
3	Create n-gram input sequences from each line of text
4	Pad all sequences to a fixed length
5	Split data into predictors and labels
6	Convert labels into categorical vectors
7	Train an LSTM RNN model
8	Save the trained model and tokenizer for inference

🏗️ Model Architecture

┌─────────────────────────────────────────┐
│       Embedding(total_words, 100)       │  ← Word vectors
├─────────────────────────────────────────┤
│    LSTM(150, return_sequences=True)     │  ← Sequence layer 1
├─────────────────────────────────────────┤
│             Dropout(0.2)               │  ← Regularization
├─────────────────────────────────────────┤
│              LSTM(100)                  │  ← Sequence layer 2
├─────────────────────────────────────────┤
│   Dense(total_words, activation=       │
│              "softmax")                │  ← Output probabilities
└─────────────────────────────────────────┘

---

🚀 Starting Commands

1. Clone or Open the Project

cd "LSTM_RNN_imple"

2. Create a Virtual Environment

python -m venv venv

Windows with TensorFlow support:

py -3.11 -m venv venv

3. Activate the Virtual Environment

Windows:

venv\Scripts\activate

macOS/Linux:

source venv/bin/activate

4. Install Dependencies

pip install -r requirements.txt

5. Run the Streamlit App

streamlit run app.py

Then open:

http://localhost:8501

---

💡 How Prediction Works

When the user enters text:

Step	Action
1	Input text is converted into token IDs
2	Token list is trimmed if longer than model input length
3	Sequence is padded
4	LSTM model predicts probabilities for all known words
5	Word with highest probability is returned as the next word

Example:

┌─────────────────────────────────┐
│  Input  : to be or not to      │
│  Output : be                   │
└─────────────────────────────────┘

---

👨‍💻 Author