Interactive development in Workspaces
This version of the Anyscale docs is deprecated. Go to the latest version for up to date information.
Open source Ray users often use Ray Jobs and Ray Serve Deployments to interactively develop their applications. With Anyscale, Workspace makes interactive development even simpler because you no longer need to package your code as a job but can run the Python script directly in Workspace. The experience becomes as easy as running the script on a local laptop, while the underlying compute can scale to hundreds or thousands of machines.
python train_model.py --cycle_lr --large_batch
Workspace makes sure the files in your project directory (/home/ray/<your_project_name>) are snapshot and replicated to all worker nodes via the runtime_env infrastructure. In contrast to open source Ray, you don't need to worry about configuring runtime_env. Simply use ray.init() to initialize your Ray app in your code.
View the Develop and Deploy Anyscale Services tutorial for more info about how to develop Serve Applications with Workspaces.
Ray Jobs APIs are not supported in the Workspace at the moment.