teachless (discontinued, use Definers directly)

Open source software for model training and prediction.

teachless provides a simple web interface to train and retrain machine learning models using local files, remote URLs, or datasets from Hugging Face. You can also use the trained models for predictions.

Features

• Flexible Training: Train models using local feature/label files or from remote sources like Hugging Face.
• Re-training: Upload an existing model to continue training with new data.
• Data Handling: Specify columns to drop, define label columns, and select specific rows or ranges for training.
• Prediction Interface: A simple tab to upload your model and data to get predictions.
• Web UI: Built with Gradio for an easy-to-use interface.

Installation

pip install "definers[cuda] @ git+https://github.com/YaronKoresh/definers.git" --extra-index-url https://pypi.nvidia.com
pip install git+https://github.com/YaronKoresh/teachless.git

This will also install the required dependencies, such as gradio and definers.

Usage

Once installed, you can launch the Gradio web interface by running the following command in your terminal:

teachless

This will start a local web server, and you can access the user interface by navigating to the provided URL.

License

This project is licensed under the MIT License. See the LICENSE file for details.

Contributing & Issues

Contributions are welcome! For bug reports and feature requests, please open an issue on the Bug Tracker.