As a machine learning engineer, I want to know when I am using a workable version of the code base so that I can manage my expectations about using this tool.
When we are ready to release the first version of this code base (and deploy it as a JAR or whatever is the right deployment format), we should tag it as v1.0.0.
As a machine learning engineer, I want to know when I am using a workable version of the code base so that I can manage my expectations about using this tool.
When we are ready to release the first version of this code base (and deploy it as a JAR or whatever is the right deployment format), we should tag it as v1.0.0.