an universal deep learning framework for training model in either distributed data parallel mode or single gpu mode

pip install .

an example experiment has been given by config/cfg_cifar100.py.

• try to use logger_info to print message, because it will write info to the log file