Coach 1.0.0 Is Now Available

The new additions to Coach take it beyond simulation-based learning environments and fuse newer and more powerful RL algorithms.

Recently, the chip giant, Intel announced the 1.0.0 release of RL Coach, which is an open-source framework for training and evaluating reinforcement learning agents.
Source: Intel 
The company said that the new release features the implementation of as many as 27 new algorithms and support for Batch Reinforcement Learning. Aso, the new release comes with improved documentation, bug fixes and several new APIs that will enable the use of Coach as a Python library.
With this release, according to the company, the main software structure of the framework has matured and stabilized and there are no major API changes on the horizon.
Intel has added support for batch reinforcement learning in Coach, while also enabling off-policy evaluation of the learned policy based on data that was acquired using another policy.
The company said that it has added several off-policy evaluators for contextual bandits and for reinforcement learning. And at the same time, it also allows the use of a wide range of integrated off-policy RL algorithms in Coach. Intel has also added support for a variant of the Batch Constrained Q Learning algorithm for discrete action space problems.