Parent Class: vizseq.scorer.VizSeqScorer
All scorers are inherited from this class and implement the
See here for the list of built-in scorers.
- corpus_level: bool = True: calculating corpus-level score. Default to
- sent_level: bool = False: calculating sentence-level score. Default to
- n_workers: Optional[int] = None: the number of processes to be used.
Nonefor automatic configuration. Default to
- verbose: bool = False: printing detailed logs.
- extra_args: Optional[Dict[str, str]] = None: a key-value dictionary for metric-specific parameters. Default to
- hypothesis: List[str]: The model predictions.
- references: List[List[str]]: The references. Inner list #1 for the 1st set of references, #2 for the 2nd set of references, and so on. (See also the example.)
- tags: Optional[List[List[str]]] = None: The sentence tags. Inner list #1 for the 1st set of tags, #2 for the 2nd
set of tags, and so on. Default to
- score: vizseq.scorers.VizSeqScore:
VizSeqScoreis a namedtuple, whose definition is as follows:
class VizSeqScore(NamedTuple):corpus_score: Optional[float] = Nonesent_scores: Optional[List[float]] = Nonegroup_scores: Optional[Dict[str, float]] = None