tensorflow/models · error · ValueError

Only support dictionary decoder_output.

Error message

Only support dictionary decoder_output.

What it means

Error "Only support dictionary decoder_output." thrown in tensorflow/models.

Source

Thrown at official/vision/modeling/heads/segmentation_heads.py:447

                                               'deeplabv3plus_sum_to_merge'}:
      # deeplabv3+ feature fusion
      x = decoder_output[str(self._config_dict['level'])] if isinstance(
          decoder_output, dict) else decoder_output
      y = backbone_output[str(self._config_dict['low_level'])] if isinstance(
          backbone_output, dict) else backbone_output
      y = self._dlv3p_norm(self._dlv3p_conv(y))
      y = self._activation(y)

      x = tf.image.resize(
          x, tf.shape(y)[1:3], method=tf.image.ResizeMethod.BILINEAR)
      x = tf.cast(x, dtype=y.dtype)
      if self._config_dict['feature_fusion'] == 'deeplabv3plus':
        x = tf.concat([x, y], axis=self._bn_axis)
      else:
        x = tf_keras.layers.Add()([x, y])
    elif self._config_dict['feature_fusion'] == 'pyramid_fusion':
      if not isinstance(decoder_output, dict):
        raise ValueError('Only support dictionary decoder_output.')
      x = nn_layers.pyramid_feature_fusion(decoder_output,
                                           self._config_dict['level'])
    elif self._config_dict['feature_fusion'] == 'panoptic_fpn_fusion':
      x = self._panoptic_fpn_fusion(decoder_output)
    else:
      x = decoder_output[str(self._config_dict['level'])] if isinstance(
          decoder_output, dict) else decoder_output

    for conv, norm in zip(self._convs, self._norms):
      x = conv(x)
      x = norm(x)
      x = self._activation(x)
    if self._config_dict['upsample_factor'] > 1:
      x = spatial_transform_ops.nearest_upsampling(
          x, scale=self._config_dict['upsample_factor'])

    return self._classifier(x)

View on GitHub (pinned to e006f5f0d5)

Solutions

1. Pass decoder_output as a dict mapping level names to feature tensors.
2. Use the standard decoder (e.g. FPN/ASPP) output instead of a plain tensor or list.

When it happens

Trigger: Thrown at official/vision/modeling/heads/segmentation_heads.py:447 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/0e8b1c28ca2a570b. Report an issue: GitHub.