tensorflow/models · error · ValueError

The output of the given metric must either be a `tf.Tensor`

Error message

The output of the given metric must either be a `tf.Tensor` or a `dict[str, tf.Tensor]`, but got unsupported output: {metric_result}.

What it means

Error "The output of the given metric must either be a `tf.Tensor` or a `dict[str, tf.Tensor]`, but got unsupported output: {metric_result}." thrown in tensorflow/models.

Source

Thrown at official/recommendation/uplift/metrics/sliced_metric.py:190

    slice_results = [metric.result() for metric in self._sliced_metrics]

    if isinstance(metric_result, tf.Tensor):
      results = {metric_name: metric_result}
      slice_names = (f"{metric_name}/{name}" for name in self._slice_names)
      results.update(zip(slice_names, slice_results))
      return results

    if isinstance(metric_result, dict) and all(
        isinstance(result, tf.Tensor) for result in metric_result.values()
    ):
      results = {**metric_result}
      for slice_name, slice_result in zip(self._slice_names, slice_results):
        result_names, result_values = zip(*slice_result.items())
        slice_names = [f"{name}/{slice_name}" for name in result_names]
        results.update(zip(slice_names, result_values))
      return results

    raise ValueError(
        "The output of the given metric must either be a `tf.Tensor` or "
        "a `dict[str, tf.Tensor]`, but got unsupported output: "
        f"{metric_result}."
    )

  def reset_state(self):
    self._metric.reset_state()
    for metric in self._sliced_metrics:
      metric.reset_state()

  def get_config(self):
    return {
        "name": self.name,
        "metric": tf_keras.metrics.serialize(self._metric),
        "slicing_spec": dict(zip(self._slice_names, self._slicing_values)),
        "slicing_feature_dtype": self._slicing_feature_dtype.name,
    }

View on GitHub (pinned to e006f5f0d5)

When it happens

Trigger: Thrown at official/recommendation/uplift/metrics/sliced_metric.py:190 when the library encounters an invalid state.

Common situations: See trigger scenarios.

AI-assisted analysis of tensorflow/models@e006f5f0d5 (2026-08-24). Data as JSON: /api/errors/76cede0501ebdc82. Report an issue: GitHub.