generate_synthetic_dataset.py

import config
from api_wrappers import hf_data_loader
from generation_steps import synthetic_end_to_start, synthetic_start_to_end, metrics_analysis


def run():
    df = hf_data_loader.load_processed_rewriting_as_pandas()

    df = synthetic_end_to_start.transform(df)
    df = synthetic_start_to_end.transform(df)
    df = metrics_analysis.transform(df)

    df.to_csv(config.SYNTHETIC_DATASET_ARTIFACT)


if __name__ == '__main__':
    run()