tensorflow/models · error · ValueError

At least one of inputs and dense_inputs must not be None.

Error message

At least one of inputs and dense_inputs must not be None.

What it means

Error "At least one of inputs and dense_inputs must not be None." thrown in tensorflow/models.

Source

Thrown at official/nlp/modeling/models/t5.py:1167

      dense_inputs: dense input data. Concat after the embedding if word ids are
        provided.
      training: whether it is training pass, affecting dropouts.

    Returns:
      output of a transformer encoder.
    """
    # Casts inputs to the dtype.
    if encoder_mask is not None:
      encoder_mask = tf.cast(encoder_mask, self.compute_dtype)
    cfg = self.config
    inputs_array = []
    if inputs is not None:
      inputs_array.append(
          self.input_embed(inputs, one_hot=cfg.one_hot_embedding))  # pyrefly: ignore[not-callable]
    if dense_inputs is not None:
      inputs_array.append(dense_inputs)
    if not inputs_array:
      raise ValueError("At least one of inputs and dense_inputs must not be "
                       "None.")
    x = tf.concat(inputs_array, axis=1)
    tensor_shape = tf_utils.get_shape_list(x)
    tensor_shape[-2] = 1
    x = self.input_dropout(x, noise_shape=tensor_shape, training=training)
    if inputs is not None:
      input_length = tf_utils.get_shape_list(inputs)[1]
    else:
      input_length = 0

    attention_outputs = []
    for i in range(cfg.num_layers):
      position_bias = self.get_relpos_bias(input_length, dense_inputs, i)
      x = self.encoder_layers[i](
          x,
          attention_mask=encoder_mask,
          position_bias=position_bias,
          training=training)

View on GitHub (pinned to e006f5f0d5)

When it happens

Trigger: Thrown at official/nlp/modeling/models/t5.py:1167 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/6f97384abbf738f9. Report an issue: GitHub.