What Does `path` do in `TabularDataLoaders.from_csv()`?

I couldn’t find it in the documentation, but I see that we define where to download the csv from, but then afterwards we also define a path parameter as well.

What’s this used for? I couldn’t necessarily decipher it from the source.

dls = TabularDataLoaders.from_csv(
    './ml_data/final_data.csv',
    path="./ml_data",
    y_names="override",
    cat_names=cat_names,
    cont_names=cont_names,
    procs=[Categorify, FillMissing, Normalize],
)