Japanese AI startup Preferred Networks (PFN) is moving ChainerRL to the PyTorch ecosystem, it was announced yesterday. The resulting PyTorch-based open-source deep Reinforcement Learning (RL) library, Preferred RL (PFRL), represents the latest in PFN’s continuing efforts to build and maintain strong ties with the PyTorch developer community.
ChainerRL is a PFN deep RL library that implements various state-of-the-art algorithms in Python using the flexible deep learning framework Chainer, which PFN open-sourced in June 2015. Last December, the company unveiled a plan to migrate its deep learning research platform from Chainer to PyTorch, and promised to provide documentation and a library for Chainer users to facilitate the transition. PFRL is the PyTorch-based successor to ChainerRL.
The main goals of PFRL are to enable reproducible research, to support a comprehensive set of algorithms and features, and to be modular and flexible, Preferred Network RL Engineer Prabhat Nagarajan wrote in a blog post. The PyTorch team tweeted that PFRL will enable users to choose amongst the library’s various algorithms and combine them with a multitude of features to produce novel RL systems.
Aiming to be both flexible and comprehensive, PFRL supports additional algorithms, including Persistent Advantage Learning, C51, ACER, A2C, and REINFORCE; and supports features that include Noisy networks, Dueling networks, Prioritized Experience Replay, Normalized Advantage Functions, and recurrent network support for most agents. These features can be easily integrated with PFRL’s algorithms.
Example scripts utilizing these algorithms and features can be found in the project GitHub.
Reporter: Yuan Yuan | Editor: Michael Sarazen
This report offers a look at how China has leveraged artificial intelligence technologies in the battle against COVID-19. It is also available on Amazon Kindle. Along with this report, we also introduced a database covering additional 1428 artificial intelligence solutions from 12 pandemic scenarios.
Click here to find more reports from us.
We know you don’t want to miss any news or research breakthroughs. Subscribe to our popular newsletter Synced Global AI Weekly to get weekly AI updates.