usage: happy-splitter [-h] -b BASE_FOLDER [-r NUM_REPEATS] [-f NUM_FOLDS]
[-t TRAIN_PERCENT] [-v VALIDATION_PERCENT] [-R]
[-H HOLDOUT_PERCENT] -o OUTPUT_FILE [-S SEED]
Generate train/validation/test splits for Happy data.
optional arguments:
-h, --help show this help message and exit
-b BASE_FOLDER, --base_folder BASE_FOLDER
Path to the Happy base folder (default: None)
-r NUM_REPEATS, --num_repeats NUM_REPEATS
Number of repeats (default: 1)
-f NUM_FOLDS, --num_folds NUM_FOLDS
Number of folds (default: 1)
-t TRAIN_PERCENT, --train_percent TRAIN_PERCENT
Percentage of data in the training set (default: 70.0)
-v VALIDATION_PERCENT, --validation_percent VALIDATION_PERCENT
Percentage of data in the validation set (default:
10.0)
-R, --use_regions Use regions in generating splits (default: False)
-H HOLDOUT_PERCENT, --holdout_percent HOLDOUT_PERCENT
Percentage of data to hold out as a holdout set
(default: None)
-o OUTPUT_FILE, --output_file OUTPUT_FILE
Path to the output split file (default:
output_split.json)
-S SEED, --seed SEED The seed to use for reproducible results (default:
None)