tfr.keras.metrics.get

Factory method to get a list of ranking metrics.

tfr.keras.metrics.get(
    key: str,
    name: Optional[str] = None,
    dtype: Optional[tf.dtypes.DType] = None,
    topn: Optional[int] = None,
    **kwargs
) -> tf.keras.metrics.Metric

Example Usage:

  metric = tfr.keras.metics.get(tfr.keras.metrics.RankingMetricKey.MRR)

to get Mean Reciprocal Rank.

  metric = tfr.keras.metics.get(tfr.keras.metrics.RankingMetricKey.MRR,
                                topn=2)

to get MRR@2.

key An attribute of RankingMetricKey, defining which metric objects to return.
name Name of metrics.
dtype Dtype of the metrics.
topn Cutoff of how many items are considered in the metric.
**kwargs Keyword arguments for the metric object.

A tf.keras.metrics.Metric. See _RankingMetric signature for more details.

ValueError If key is unsupported.