Train multi-task image, text, or ensemble (image + text) models - View it on GitHub