model-logger-dp 1.0.3.post2

Model Logger for Pytorch

Model Logger

A package that contains method to log the model training log.

The package is an uncompleted package in the version 1.x. Its functions will be gradually improved in subsequent versions.

How to use

To use the model logger, you need to install the package first. You can install it using pip:

pip install model_logger_dp

To start the model logger.

import os
import model_logger_dp

os.environ['LOG_PATH'] = './output/log'  # Set the directory for logs

# If you want to save the log with a specific file name, you should set the filename parameter.
# If you do not set the filename parameter, the log will be saved with a datatime name.
logger  = model_logger_dp.ModelLogger(filename='train.log')

# Use the print method to log the message.
print('Hellow World!')

To look the help and update log of the package, you can use the log_help and log_update method.

import model_logger_dp

model_logger_dp.log_help()
model_logger_dp.log_update()

Update

`1.0.3` - Add model summery method and update the setup method.
`1.0.2` - Add showing the caller method.
`1.0.1` - Add the logger decorator, improve the logger instance and implement help and update log method.
`1.0.0` - Initial release with basic logging functionality.

License

model_logger is MIT licensed. See the LICENSE for details.