aac_metrics.eval module

load_csv_file(
fpath: str | Path,
cands_columns: str | Iterable[str] = ('caption_predicted',),
mrefs_columns: str | Iterable[str] = ('caption_1', 'caption_2', 'caption_3', 'caption_4', 'caption_5'),
load_mult_cands: bool = False,
strict: bool = True,
) → tuple[list, list[list[str]]][source]

Load candidates and mult_references from a CSV file.

Parameters:

  • fpath – The filepath to the CSV file.

  • cands_columns – The columns of the candidates. defaults to (“captions_predicted”,).

  • mrefs_columns – The columns of the multiple references. defaults to (“caption_1”, “caption_2”, “caption_3”, “caption_4”, “caption_5”).

  • load_mult_cands – If True, load multiple candidates from file. defaults to False.

Returns:

A tuple of (candidates, mult_references) loaded from file.