It would be useful for most R&D tasks to be able to train a model using data stored as csv files on labtenant instead of fetching data from our database for every new run.
We could add a new options --train_from_file <filepath>, --test_from_file <filepath> and --predict_from_file <filepath> and create a function main.fetch_data(), which would:
load data from CSV files if those three arguments exists in the argparse.Namespace
It would be useful for most R&D tasks to be able to train a model using data stored as
csvfiles on labtenant instead of fetching data from our database for every new run.We could add a new options
--train_from_file <filepath>,--test_from_file <filepath>and--predict_from_file <filepath>and create a function main.fetch_data(), which would: