General good practice for logging in python code is to pass all messages through the central logging module, and provide handlers as outputs, which are often a StreamHandler to go to the console and a FileHandler to go to a text file. It would be good to extend file handler so that at the end of a run, the file contain the logs is logged with MLflow as an artifact. My invesatigations show that ML Flow does not capture stdout/stderr by default, and this would be a good way to continue python best practice into ML projects
General good practice for logging in python code is to pass all messages through the central logging module, and provide handlers as outputs, which are often a StreamHandler to go to the console and a FileHandler to go to a text file. It would be good to extend file handler so that at the end of a run, the file contain the logs is logged with MLflow as an artifact. My invesatigations show that ML Flow does not capture stdout/stderr by default, and this would be a good way to continue python best practice into ML projects