keras-team/keras · error · ValueError

The argument `strides` cannot contains 0. Received strides={

Error message

The argument `strides` cannot contains 0. Received strides={self.strides}

What it means

Depthwise convolution strides must be positive in every spatial dimension; a 0 stride is meaningless for sliding a filter. The constructor rejects any strides tuple containing 0 (the message contains the grammar typo 'cannot contains').

Source

Thrown at keras/src/layers/convolutional/base_depthwise_conv.py:145

        self.bias_constraint = constraints.get(bias_constraint)
        self.input_spec = InputSpec(min_ndim=self.rank + 2)
        self.data_format = self.data_format

        if self.depth_multiplier is not None and self.depth_multiplier <= 0:
            raise ValueError(
                "Invalid value for argument `depth_multiplier`. Expected a "
                "strictly positive value. Received "
                f"depth_multiplier={self.depth_multiplier}."
            )

        if not all(self.kernel_size):
            raise ValueError(
                "The argument `kernel_size` cannot contain 0. Received "
                f"kernel_size={self.kernel_size}."
            )

        if not all(self.strides):
            raise ValueError(
                "The argument `strides` cannot contains 0. Received "
                f"strides={self.strides}"
            )

    def build(self, input_shape):
        if self.data_format == "channels_last":
            channel_axis = -1
            input_channel = input_shape[-1]
        else:
            channel_axis = 1
            input_channel = input_shape[1]
        self.input_spec = InputSpec(
            min_ndim=self.rank + 2, axes={channel_axis: input_channel}
        )
        self.compute_output_shape(input_shape)
        depthwise_shape = self.kernel_size + (
            input_channel,
            self.depth_multiplier,

View on GitHub (pinned to 7a34a03db6)

Solutions

1. Set strides to positive integers in every dimension (default 1).
2. Validate all(s > 0 for s in strides) before layer creation.
3. Prefer strides=1 with pooling when downsampling intent is unclear.

Example fix

# before
keras.layers.DepthwiseConv2D(3, strides=(2, 0))

# after
keras.layers.DepthwiseConv2D(3, strides=(2, 2))

st = strides if isinstance(strides, (tuple, list)) else (strides,)
assert all(s > 0 for s in st)

def valid_strides(strides) -> bool:
    st = strides if isinstance(strides, (tuple, list)) else (strides,)
    return all(s > 0 for s in st)

When it happens

Trigger: Constructing DepthwiseConv1D/2D/3D with strides containing a 0, e.g. strides=(2,0), often from a computed downsampling factor.

Common situations: Strides derived from arithmetic on input dims (stride = size // factor underflowing to 0); mis-pasting a kernel_size tuple into strides.

Related errors

• Invalid value for argument `depth_multiplier`. Expected a st
• The argument `kernel_size` cannot contain 0. Received kernel
• The argument `strides` cannot contains 0(s). Received: strid
• If using `weights="imagenet"` with `include_top=True`, `clas
• `strides > 1` not supported in conjunction with `dilation_ra

AI-assisted analysis of keras-team/keras@7a34a03db6 (2026-08-25). Data as JSON: /api/errors/a10dc83eb0a36672. Report an issue: GitHub.