How much python workers should be created per executor

Question

I\'m working on a pipeline that inferences ml model using ML Flow and PySpark (2.4.0).

I have read the whole docs and articles about inter-process communications (for exe