define train, test and validation splits by redefining id_start and id_end in the loading script (Lenze_dataset.py)
you should define a config_name when calling the function load_dataset. the config_name is the name of a folder like H045_I_ccw_withoutLoad
example: dataset = load_dataset("Lenze_dataset.py","H045_I_ccw_withoutLoad",trust_remote_code=True)