Basic Training Instructions
pip install -r requirements
python main.py --phase train --exp "test_exp" --nb_epochs 150
This should create a chekpoint directory under ckdir/<modelID>/
where is the ID
of the model created during training.
Basic Testing Instructions
python main.py --phase test --model_id <modelID>
This should evaluate the generated model against the test dataset.
Basic Exporting Instructions
python export.py --model_id <modelID> --target saved_model
This exports the trained model for inference using saved_model method of tensorflow. It also tests the
model against the test dataset. The exported model ready to be used with tf-serving is located at
export_dir/<modelID>/serving
Help
python main.py --h