Logging

Callbacks for logging argus model training process.

class argus.callbacks.LoggingToFile(file_path, create_dir=True, formatter='[%(asctime)s][%(levelname)s]: %(message)s', append=False)

Write the argus model training progress into a file.

It adds a standard Python logger to log all losses and metrics values during training. The logger is used to output other messages, like info from callbacks and errors.

Parameters

• file_path (str) – Path to the logging file.

• create_dir (bool, optional) – Create the directory for the logging file if it does not exist. Defaults to True.

• formatter (str, optional) – Standard Python logging formatter to format the log messages. Defaults to ‘%(asctime)s %(levelname)s %(message)s’.

• append (bool, optional) – Append the log file if it already exists or rewrite it. Defaults to False.

class argus.callbacks.LoggingToCSV(file_path, create_dir=True, separator=',', write_header=True, append=False)

Write the argus model training progress into a CSV file.

It logs all losses and metrics values during training into a .csv file for for further analysis or visualization.

Parameters

• file_path (str) – Path to the .csv logging file.

• create_dir (bool, optional) – Create the directory for the logging file if it does not exist. Defaults to True.

• separator (str, optional) – Values separator character to use. Defaults to ‘,’.

• write_header (bool, optional) – Write the column headers. Defaults to True.

• append (bool, optional) – Append the log file if it already exists or rewrite it. Defaults to False.