all launch files in directory bin all configs in configuration dir for each config we have experiment name - directory with its name creating in data directory and there we store all data
lets see how we can create something
you can create your own model in pipeline.models and extends from ModelBase class in pipeline.models.base and do all in abstract methods
you can create your own data builder in pipeline.data ans extends from DataBuilderBase class in pipeline.data.base
after you should add your class in configuration
and you should put train.csv and test.csv in data/{experiment_name} folder
after you can launch something like
PYTHONPATH=. python3 bin/{process} /path/to/config
or export PYTHONPATH=.
and after use without it
for example if we want build training data so we should enter this
python3 bin/build_training_data.py configuration/base.py
info and higher logs write to console all application logs write to debug.log and all logs write to all_debug.log
you can watch it like tail -f debug.log