tensorflow/models · error · ValueError

"likelihood" must be one of {_SUPPORTED_LIKELIHOOD}, got {li

Error message

"likelihood" must be one of {_SUPPORTED_LIKELIHOOD}, got {likelihood}.

What it means

Error ""likelihood" must be one of {_SUPPORTED_LIKELIHOOD}, got {likelihood}." thrown in tensorflow/models.

Source

Thrown at official/nlp/modeling/layers/gaussian_process.py:299

      covariance matrix by passing through data once (say in the final epoch).
    ridge_penalty: (float) Initial Ridge penalty to weight covariance matrix.
      This value is used to stablize the eigenvalues of weight covariance
      estimate so that the matrix inverse can be computed for Cov = inv(t(X) * X
      + s * I). The ridge factor s cannot be too large since otherwise it will
      dominate the t(X) * X term and make covariance estimate not meaningful.
    likelihood: (str) The likelihood to use for computing Laplace approximation
      for the covariance matrix. Can be one of ('binary_logistic', 'poisson',
      'gaussian').
  """

  def __init__(self,
               momentum=0.999,
               ridge_penalty=1.,
               likelihood='gaussian',
               dtype=None,
               name='laplace_covariance'):
    if likelihood not in _SUPPORTED_LIKELIHOOD:
      raise ValueError(
          f'"likelihood" must be one of {_SUPPORTED_LIKELIHOOD}, got {likelihood}.'
      )
    self.ridge_penalty = ridge_penalty
    self.momentum = momentum
    self.likelihood = likelihood
    super(LaplaceRandomFeatureCovariance, self).__init__(dtype=dtype, name=name)

  def compute_output_shape(self, input_shape):
    gp_feature_dim = input_shape[-1]
    return tf.TensorShape([gp_feature_dim, gp_feature_dim])

  def build(self, input_shape):
    gp_feature_dim = input_shape[-1]

    # Convert gp_feature_dim to int value for TF1 compatibility.
    if isinstance(gp_feature_dim, tf.compat.v1.Dimension):
      gp_feature_dim = gp_feature_dim.value

View on GitHub (pinned to e006f5f0d5)

When it happens

Trigger: Thrown at official/nlp/modeling/layers/gaussian_process.py:299 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/8656dd64aa05db9e. Report an issue: GitHub.