tensorflow/models · error · ValueError

If `sequence_output` is not in encoder output,`latent_output

Error message

If `sequence_output` is not in encoder output,`latent_output` must be in encoder output.

What it means

Error "If `sequence_output` is not in encoder output,`latent_output` must be in encoder output." thrown in tensorflow/models.

Source

Thrown at official/projects/perceiver/modeling/models/pretrainer.py:177

        'input_mask': input_mask,
        'input_type_ids': input_type_ids,
    }
    encoder_outputs = self.encoder(encoder_inputs)

    if 'sequence_output' not in encoder_outputs:
      if 'latent_output' in encoder_outputs:
        z = encoder_outputs['latent_output']
        decoder_inputs = {'latent_output': z, 'input_mask': input_mask}
        decoder_output = self._decoder(decoder_inputs)  # pyrefly: ignore[not-callable]

        outputs = dict()
        if isinstance(decoder_output, dict):
          outputs = decoder_output
        else:
          raise ValueError('decoder\'s output should be a dict,'
                           f'but got {decoder_output}')
      else:
        raise ValueError('If `sequence_output` is not in encoder output,'
                         '`latent_output` must be in encoder output.')
    else:
      outputs = encoder_outputs

    sequence_output = outputs['sequence_output']
    # Inference may not have masked_lm_positions and mlm_logits is not needed.
    if 'masked_lm_positions' in inputs:
      masked_lm_positions = inputs['masked_lm_positions']
      outputs['mlm_logits'] = self.masked_lm(
          sequence_output, masked_positions=masked_lm_positions)
    return outputs

  @property
  def checkpoint_items(self):
    """Returns a dictionary of items to be additionally checkpointed."""
    items = dict(
        encoder=self.encoder,
        masked_lm=self.masked_lm,

View on GitHub (pinned to e006f5f0d5)

When it happens

Trigger: Thrown at official/projects/perceiver/modeling/models/pretrainer.py:177 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/a947ba84437dfde0. Report an issue: GitHub.