Usage of the Logger

Using the Default Logger of AcuiRT

Create a Logger

• When using the default logger, you need to create an instance of the AcuiRTDefaultLogger class from the acuirt.logger module.

• Specify name and set the logger’s name.

  from acuirt.logger import AcuiRTDefaultLogger
  logger = AcuiRTDefaultLogger(name="AcuiRT")

Apply the logger

• You can pass a logger to the following function.

  summary = convert_model(resnet, config, path, False, data, logger=logger)
  model = load_runtime_modules(resnet, summary, path, logger=logger)

Change the Minimum Log Level

• To change the minimum log level, set min_severity in the argument.

• The correspondence table of log levels is as follows.

  TensorRT ILogger Log Level	Python logging Log Level
  ILogger.INTERNAL_ERROR	logging.CRITICAL
  ILogger.ERROR	logging.ERROR
  ILogger.WARNING	logging.WARNING
  ILogger.INFO	logging.INFO
  ILogger.VERBOSE	logging.DEBUG

Change the output format

• If you want to change the output format, please set it using the set_formatter() function.

  logger = AcuiRTDefaultLogger()
  formatter = logging.Formatter(
      "[%(asctime)s]: %(levelname)s: %(message)s",
      )
  logger.set_formatter(formatter)

Add Output Destination

• If you want to add an output destination, set it from the set_handler() function.

• If you also want to specify the output format, set the formatter argument.

  import sys
  handler = logging.StreamHandler(sys.stdout)
  logger.set_handler(handler)

Use a custom logger

• The intelligence/acuirt/use_custom_logger.py in aibooster-examples shows an example of applying a custom logger to AcuiRT.