strides > 1 requires use_projections=True, otherwise the inp

Error message

strides > 1 requires use_projections=True, otherwise the inputs and shortcut will have shape mismatch

What it means

Error "strides > 1 requires use_projections=True, otherwise the inputs and shortcut will have shape mismatch" thrown in tensorflow/models.

Source

Thrown at official/projects/assemblenet/modeling/assemblenet.py:266

    inter_filters: `int` number of filters for the second conv. layer.
    strides: `int` block stride. If greater than 1, this block will ultimately
      downsample the input spatially.
    use_projection: `bool` for whether this block should use a projection
      shortcut (versus the default identity shortcut). This is usually `True`
      for the first block of a block group, which may change the number of
      filters and the resolution.
    num_frames: `int` number of frames in the input tensor.
    temporal_dilation: `int` temporal dilatioin size for the 1D conv.
    bn_decay: `float` batch norm decay parameter to use.
    bn_epsilon: `float` batch norm epsilon parameter to use.
    use_sync_bn: use synchronized batch norm for TPU.
    step: `int` to decide whether to put 2D module or (2+1)D module.

  Returns:
    The output `Tensor` of the block.
  """
  if strides > 1 and not use_projection:
    raise ValueError('strides > 1 requires use_projections=True, otherwise the '
                     'inputs and shortcut will have shape mismatch')
  shortcut = inputs
  if use_projection:
    # Projection shortcut only in first block within a group. Bottleneck blocks
    # end with 4 times the number of filters.
    filters_out = 4 * filters
    shortcut = conv2d_fixed_padding(
        inputs=inputs, filters=filters_out, kernel_size=1, strides=strides)
    shortcut = rf.build_batch_norm(
        bn_decay=bn_decay, bn_epsilon=bn_epsilon, use_sync_bn=use_sync_bn)(
            shortcut)

  if step % 2 == 1:
    k = 3

    inputs = reshape_temporal_conv1d_bn(
        inputs=inputs,
        filters=filters,

View on GitHub (pinned to e006f5f0d5)

When it happens

Trigger: Thrown at official/projects/assemblenet/modeling/assemblenet.py:266 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/5c2ff8195e752c29. Report an issue: GitHub.