Description

neon is Intel Nervana's python-based deep learning framework designed for data scientists and software engineers. It enables users to optimize deep learning model performance on all hardware, define data models in Python, access database application programming interfaces (APIs), create and run tables and migrations in the database, generate forms from existing models, validate user-submitted data, automatically acquire and release database resources, as well as specify the translation and formatting part of the application. Features include batch prediction, Rectlin activation function, macrobatching DataLoader, array iterator, and dataset pre-processing.