Azure Machine Learning reinforcement learning via the azureml.contrib.train.rl package will be retired on 30 June 2022. Before that date, please use the Ray on Azure Machine Learning Library for reinforcement learning experiments with Azure Machine Learning. For example, see the notebook Reinforcement Learning in Azure Machine Learning - Pong problem.
Ray on Azure Machine Learning Library gives you the same level of integration for reinforcement learning experiments plus enhancements such as:
- Ability to simplify cluster management.
- Ease of converting Azure Machine Learning compute instance and compute cluster into Ray and Dask cluster.
- Ability to complete all the complex wirings and setup of a Ray cluster with Dask behind the scene while exposing a simple Ray context object for users perform parallel Python computing tasks.
Required action
To avoid service disruptions, update the experiments and pipelines using azureml.contrib.train.rl and update them to follow a similar pattern as this notebook Reinforcement Learning in Azure Machine Learning - Pong problem.
Help and support
If you have a support plan and you need technical help, please create a support request.