preprocess.py

from datasets import load_dataset
from config import DATASET_EN_ES

def load_and_prepare():
    dataset = load_dataset(DATASET_EN_ES)

    def format_example(example):
        return {
            "source": example["term"]["en"],
            "target": example["term"]["es"]
        }

    dataset = dataset.map(format_example)
    return dataset

if __name__ == "__main__":
    ds = load_and_prepare()
    print(ds["train"][0])