# ONE (On-device Neural Engine)

A high-performance on-device neural network inference framework from Samsung, with a runtime for CPUs, GPUs, DSPs, and NPUs and a compiler toolchain that converts models trained in frameworks such as TensorFlow or PyTorch into a unified form.

ONE, short for On-device Neural Engine, is a neural network inference framework aimed at running trained models efficiently on-device across processor types including CPUs, GPUs, DSPs, and NPUs. It provides a runtime that targets Linux kernel-based platforms such as Ubuntu, Tizen, and Android, alongside a compiler toolchain that converts models built with training frameworks such as TensorFlow or PyTorch into a single, unified form the runtime can execute.

Feature development is guided partly by community feature requests, which are tracked and prioritized by upvotes on a public list, and contributors are encouraged to implement operator kernels themselves using the project's guide for adding a new operation.

It is written mainly in Python.

Website: https://github.com/Samsung/ONE