tensorflow/models · error · ValueError

Number of input channels: {input_shape[-1]} are not divisibl

Error message

Number of input channels: {input_shape[-1]} are not divisible by number of groups: {self._groups}.

What it means

Error "Number of input channels: {input_shape[-1]} are not divisible by number of groups: {self._groups}." thrown in tensorflow/models.

Source

Thrown at official/projects/edgetpu/vision/modeling/custom_layers.py:353

      raise ValueError('Valid padding options are : same, or valid.')

    self._groups = groups
    for _ in range(self._groups):
      # Override the activation so that batchnorm can be applied after the conv.
      self.conv_layers.append(
          tf_keras.layers.Conv2D(per_conv_filter_size, kernel_size, **kwargs))

    if self.use_batch_norm:
      for _ in range(self._groups):
        self.bn_layers.append(
            self.batch_norm_layer(  # pyrefly: ignore[not-callable]
                axis=-1, momentum=bn_momentum, epsilon=bn_epsilon))  # pytype: disable=bad-return-type  # typed-keras

  def call(self, inputs: Any) -> Any:  # pytype: disable=signature-mismatch  # overriding-parameter-count-checks
    """Applies 2d group convolution on the inputs."""
    input_shape = inputs.get_shape().as_list()
    if input_shape[-1] % self._groups != 0:
      raise ValueError(
          f'Number of input channels: {input_shape[-1]} are not divisible '
          f'by number of groups: {self._groups}.')
    input_slices = tf.split(inputs, num_or_size_splits=self._groups, axis=-1)
    output_slices = []
    for g in range(self._groups):
      output_slice = self.conv_layers[g](input_slices[g])
      if self.use_batch_norm:
        output_slice = self.bn_layers[g](output_slice)
      output_slice = self.activation(output_slice)  # pyrefly: ignore[not-callable]
      output_slices.append(output_slice)

    outputs = tf.concat(output_slices, axis=-1)
    return outputs


def _nnapi_scalar(value, dtype):
  # Resolves "Scalar operand should be constant" at cost of broadcasting
  return tf.constant(value, dtype=dtype, shape=(1,))

View on GitHub (pinned to e006f5f0d5)

When it happens

Trigger: Thrown at official/projects/edgetpu/vision/modeling/custom_layers.py:353 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/aa03de413fd3e89d. Report an issue: GitHub.