markbotros1 / neural-machine-translation

An LSTM-based Recurrent Neural Network that translates Modern English into Old (Shakespearean) English

neural-machine-translation

An LSTM-based Recurrent Neural Network that translates Modern English into Old (Shakespearean) English

Summary

• Encoder-Decoder RNN that takes modern English as input and outputs Old English
• Fronted interface built with Flask to interact with model

Running the project

1. Clone repo
2. Create and activate virtual environment

$cd path/to/NeuralMachineTranslation

$python -m venv project-env
$source project-env/bin/activate

3. Install requirements

$pip install -r requirements.txt

4. From command line, run app using following

$python src/app.py

5. Paste the generated localhost server url into browser

Example: http://127.0.0.1:5000/

Improving the model

1. Modify the RNN's architecture found in: src/nmt.py
2. Finetune the model's hyperparameters in: src/train_model.py

Project Organization

├── README.md          <- The top-level README for developers using this project.
├── datasets           <- Data files used to train and test model
|
├── models             <- Trained and serialized models
│
├── requirements.txt   <- The requirements file for reproducing the analysis environment, e.g.
│                         generated with `pip freeze > requirements.txt`
│
├── src                <- Source code for use in this project.
│   ├── __init__.py    <- Makes src a Python module
│   ├── app.py         <- Runs web application
│   │
│   ├── data           <- Helper functions to carry out data preprocessing
│   │   ├── preprocessing.py
│   │   └── vocab.py
│   │
│   ├── model          <- Modules and scripts to build/train models and make translations
│   │   ├── encoder.py
│   │   ├── decoder.py
│   │   ├── nmt.py
│   │   ├── train_model.py
│   │   └── predict_model.py
│   ├── static         <- Contains JavaScript and CSS files
│   │   ├── site.css
│   │   └── input.js   
│   ├── templates      <- Contains HTML
│   │   └── index.html