yutingye / FRL-Distributed-ML-Scaffold

Define an ML problem to train with Pytorch and to leverage Pytorch's functionality for multiprocessing and distributed compute.

FRL-Distributed-ML-Scaffold

FRL Distributed ML Scaffold is a set of training scripts intended to simplify defining, training, and debugging a multi-task machine learning problem. Problems implemented on this framework get out-of-the-box distributed training and multithreaded online data preprocessing support.

Requirements

FRL Distributed ML Scaffold requires or works with

• Mac OS X or Linux

Getting Started with FRL Distributed ML Scaffold

To get started, run setup.py install. Set up a problem by inheriting from and implementing the API from the Problem class in problem.py. The runner for problems is Solver.solve().

See the CONTRIBUTING file for how to help out.

License

FRL Distributed ML Scaffold is MIT licensed, as found in the LICENSE file.