tensorflow/models · error · ValueError

`from_logits` must be set to `True` when `y_pred` is of type

Error message

`from_logits` must be set to `True` when `y_pred` is of type TwoTowerTrainingOutputs. Note that the true logits and true predictions are assumed to be linked to each other through the log link function: `true_logits = tf.math.log(true_predictions).

What it means

Error "`from_logits` must be set to `True` when `y_pred` is of type TwoTowerTrainingOutputs. Note that the true logits and true predictions are assumed to be linked to each other through the log link function: `true_logits = tf.math.log(true_predictions)." thrown in tensorflow/models.

Source

Thrown at official/recommendation/uplift/metrics/poisson_metrics.py:107

    """
    super().__init__(
        loss_fn=tf.nn.log_poisson_loss,
        from_logits=from_logits,
        compute_full_loss=compute_full_loss,
        slice_by_treatment=slice_by_treatment,
        name=name,
        dtype=dtype,
    )

  def update_state(
      self,
      y_true: tf.Tensor,
      y_pred: types.TwoTowerTrainingOutputs | tf.Tensor,
      sample_weight: tf.Tensor | None = None,
  ):
    if not self._from_logits:
      if isinstance(y_pred, types.TwoTowerTrainingOutputs):
        raise ValueError(
            "`from_logits` must be set to `True` when `y_pred` is of type"
            " TwoTowerTrainingOutputs. Note that the true logits and true"
            " predictions are assumed to be linked to each other through the"
            " log link function: `true_logits = tf.math.log(true_predictions)."
        )
      y_pred = tf.math.log(y_pred)

    super().update_state(y_true, y_pred, sample_weight)

  def get_config(self) -> dict[str, Any]:
    config = super().get_config()
    del config["loss_fn"]
    return config

  @classmethod
  def from_config(cls, config: dict[str, Any]) -> LogLoss:
    return cls(**config)

View on GitHub (pinned to e006f5f0d5)

When it happens

Trigger: Thrown at official/recommendation/uplift/metrics/poisson_metrics.py:107 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/eb7406db2aeeec87. Report an issue: GitHub.