Hardcoded target_colname in /content/rtfm/rtfm/data_sources/utils.py

[dariabroscoteanu]

def generate_files_from_csv(
    csv_src: str, output_dir: str, to_regression: bool, task: Optional[str] = None
):
    """Generate the FeatureList and TaskConfig for a CSV file,
    and write the results (including copy of CSV) to output_dir."""
    if task is None:
        task = os.path.basename(csv_src).replace(".csv", "")

    df = pd.read_csv(csv_src)

    df.columns = [clean_colname(colname) for colname in df.columns]

    target_colname = df.columns[-1]

target_colname is required for running the quick evaluation pipeline, but it is not used in the code correctly, only if it is the last column in the dataset.