tensorflow/models · error · ValueError

TransformerDecoderBlock must have 5 inputs, when it uses mul

Error message

TransformerDecoderBlock must have 5 inputs, when it uses multi_channel_cross_attention. But it got: %d

What it means

Error "TransformerDecoderBlock must have 5 inputs, when it uses multi_channel_cross_attention. But it got: %d" thrown in tensorflow/models.

Source

Thrown at official/nlp/modeling/layers/transformer.py:401

            self._attention_initializer, use_legacy_format=True
        ),
        "self_attention_cls": self._self_attention_cls,
        "cross_attention_cls": self._cross_attention_cls,
    }
    base_config = super().get_config()
    return dict(list(base_config.items()) + list(config.items()))

  def common_layers_with_encoder(self):
    """Gets layer objects that can make a Transformer encoder block."""
    return [
        self.self_attention, self.self_attention_layer_norm,
        self.intermediate_dense, self.output_dense, self.output_layer_norm
    ]

  def call(self, inputs, cache=None, decode_loop_step=None):
    if self.multi_channel_cross_attention:
      if len(inputs) != 5:
        raise ValueError(
            "TransformerDecoderBlock must have 5 inputs, when it uses "
            "multi_channel_cross_attention. But it got: %d" % len(inputs))
    elif len(inputs) != 4:
      raise ValueError(
          "TransformerDecoderBlock must have 4 inputs, but it got: %d" %
          len(inputs))
    input_tensor, memory, attention_mask, self_attention_mask = inputs[:4]
    source_tensor = input_tensor
    if self._norm_first:
      input_tensor = self.self_attention_layer_norm(input_tensor)
    self_attention_output, cache = self.self_attention(
        query=input_tensor,
        value=input_tensor,
        attention_mask=self_attention_mask,
        cache=cache,
        decode_loop_step=decode_loop_step)
    self_attention_output = self.self_attention_dropout(self_attention_output)
    if self._norm_first:

View on GitHub (pinned to e006f5f0d5)

When it happens

Trigger: Thrown at official/nlp/modeling/layers/transformer.py:401 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/366061fc4c2d9fa8. Report an issue: GitHub.