Getting started is as easy as:
python train.py
More advanced usage can be done with the CLI flags:
python train.py --help
flags:
train.py:
--artifacts_dir: artifact save dir
--checkpoint_path: model checkpoint directory
(default: 'artifacts/checkpoint/diffusion_model')
--epochs: epochs to train for
(default: '100')
(an integer)
--[no]force_download: Whether or not to force download the dataset.
(default: 'false')
--model_dir: directory to save model to
--percent: percentage of dataset to use
(default: '100.0')
(a number)