tensorflow/models · error · ValueError

hidden_size must be a multiple of 2.

Error message

hidden_size must be a multiple of 2.

What it means

Error "hidden_size must be a multiple of 2." thrown in tensorflow/models.

Source

Thrown at official/projects/pix2seq/modeling/pix2seq_model.py:248

      top_k=0,
      top_p=0.4,
      early_stopping_token: int | None = None,
      **kwargs,
  ):
    super().__init__(**kwargs)
    self._backbones = backbones
    self._backbone_endpoint_names = backbone_endpoint_names
    self._max_seq_len = max_seq_len
    self._vocab_size = vocab_size
    self._hidden_size = hidden_size
    self._num_heads = num_heads
    self._num_encoder_layers = num_encoder_layers
    self._num_decoder_layers = num_decoder_layers
    self._drop_path = drop_path
    self._drop_units = drop_units
    self._drop_att = drop_att
    if hidden_size % 2 != 0:
      raise ValueError("hidden_size must be a multiple of 2.")
    if len(encoded_feature_dropout_rates) != len(self._backbones):
      raise ValueError(
          "The length of encoded_feature_dropout_rates must be equal to the "
          "number of backbones."
      )

    self._encoder_dropouts = [
        tf_keras.layers.Dropout(r) for r in encoded_feature_dropout_rates
    ]
    # Separate projections and learned layer normalization for each image.
    num_backbones = len(self._backbones)
    self._stem_projections = [
        tf_keras.layers.Dense(self._hidden_size, name="stem_projection")
        for _ in range(num_backbones)
    ]
    self._stem_lns = [
        tf_keras.layers.LayerNormalization(epsilon=1e-6, name="stem_ln")
        for _ in range(num_backbones)

View on GitHub (pinned to e006f5f0d5)

When it happens

Trigger: Thrown at official/projects/pix2seq/modeling/pix2seq_model.py:248 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/434a41dee75068ca. Report an issue: GitHub.