update for binary

bias_auc.py

@@ -1,3 +1,5 @@
+%%writefile test_metric/test_metric.py
+
 import evaluate
 import datasets
 from datasets.features import Sequence, Value, ClassLabel
@@ -30,7 +32,6 @@ Args:
     label list[int]: list containing label index for each item
     output list[list[float]]: list of model output values for each
     subgroup list[str] (optional): list of subgroups that appear in target to compute metric over
-
 Returns (for each subgroup in target):
     'Subgroup' : Subgroup AUC score,
     'BPSN' : BPSN (Background Positive, Subgroup Negative) AUC,
@@ -49,7 +50,6 @@ Example:
     ... [0.4341845214366913, 0.5658154487609863],
     ... [0.400595098733902, 0.5994048714637756],
     ... [0.3840397894382477, 0.6159601807594299]]
-
     >>> metric = load('Intel/bias_auc')
     >>> metric.add_batch(target=target,
                      label=label,
@@ -67,7 +67,7 @@ class BiasAUC(evaluate.Metric):
             features=datasets.Features(
                 {
                  'target': Sequence(feature=Value(dtype='string', id=None), length=-1, id=None),
-                 'label': Sequence(feature=Value(dtype='int64', id=None), length=-1, id=None),
+                 'label': Value(dtype='int64', id=None),
                  'output': Sequence(feature=Value(dtype='float32', id=None), length=-1, id=None),
                 }
             ),
@@ -78,7 +78,7 @@ class BiasAUC(evaluate.Metric):
         """Returns label and output score from `targets` and `labels`
         if `subgroup` is in list of targeted groups found in `targets`
         """
-        target_class = target_class if target_class is not None else np.asarray(outputs).shape[-1] - 1
+        target_class = target_class if target_class is not None else 0
         for target, label, result in zip(targets, labels, outputs):
             if subgroup in target:
                  yield label, result[target_class]
@@ -89,7 +89,7 @@ class BiasAUC(evaluate.Metric):
         label is not the same as `target_class`; or (2) `subgroup` is not in list of
         targeted groups found in `targets` and  label is the same as `target_class`
         """
-        target_class = target_class if target_class is not None else np.asarray(outputs).shape[-1] - 1
+        target_class = target_class if target_class is not None else 1
         for target, label, result in zip(targets, labels, outputs):
             if not target:
                 continue
@@ -107,7 +107,7 @@ class BiasAUC(evaluate.Metric):
         targeted groups found in `targets` and  label is not the same as `target_class`
         """
         # get the index from class
-        target_class = target_class if target_class is not None else np.asarray(outputs).shape[-1] - 1
+        target_class = target_class if target_class is not None else 1
         for target, label, result in zip(targets, labels, outputs):
             if not target:
                 continue