keras-team/keras · error · ValueError

Invalid data_format: {data_format}

Error message

Invalid data_format: {data_format}

What it means

UpSampling3D._resize_volumes raises when data_format matches neither 'channels_first' nor 'channels_last' after the if/elif chain. Like the 2D case, the public constructor normally validates earlier, so this fires on direct/internal calls with non-standard strings ('NCDHW', typos, None).

Source

Thrown at keras/src/layers/reshaping/up_sampling3d.py:134

            height_factor: Positive integer.
            width_factor: Positive integer.
            data_format: One of `"channels_first"`, `"channels_last"`.

        Returns:
            Resized tensor.
        """
        if data_format == "channels_first":
            output = ops.repeat(x, depth_factor, axis=2)
            output = ops.repeat(output, height_factor, axis=3)
            output = ops.repeat(output, width_factor, axis=4)
            return output
        elif data_format == "channels_last":
            output = ops.repeat(x, depth_factor, axis=1)
            output = ops.repeat(output, height_factor, axis=2)
            output = ops.repeat(output, width_factor, axis=3)
            return output
        else:
            raise ValueError(f"Invalid data_format: {data_format}")

View on GitHub (pinned to 7a34a03db6)

Solutions

1. Pass exactly 'channels_first' or 'channels_last'
2. Route custom code through keras.backend.standardize_data_format() before calling the helper, or use the public UpSampling3D layer API
3. Set data_format once via keras.config.image_data_format and omit per-layer arguments

Example fix

# before
out = up3d._resize_volumes(x, 2, 2, 2, data_format='NCDHW')

# after
out = up3d._resize_volumes(x, 2, 2, 2, data_format='channels_first')

def valid_data_format(df):
    return df in {'channels_last', 'channels_first'}

assert valid_data_format(df), f'bad data_format: {df}'

def is_keras_data_format(v) -> bool:
    return v in ('channels_last', 'channels_first')

When it happens

Trigger: Invoking _resize_volumes directly (custom subclass, monkey-patch, or copied helper code) with data_format='NCDHW', 'none', or an unset variable; a wrapper that passes the raw user string down unchecked.

Common situations: Porting PyTorch-style format names (NCDHW/NDHWC) into Keras layer code; custom 3D upsampling wrappers that accept arbitrary strings and forward them unchecked.

Understand the failure class

Background: "Must be a positive integer", "Invalid value", "Unsupported": the invalid-argument-value error family, when a library rejects the value you pass — this error's family across 35 libraries.

Related errors

• Invalid `data_format` argument: {data_format}
• Unknown data_format: {data_format}
• `data_format` should be `"channels_last"` (channel after row
• Unknown activation function '{activation}' cannot be seriali
• Could not interpret activation function identifier: {identif

AI-assisted analysis of keras-team/keras@7a34a03db6 (2026-08-25). Data as JSON: /api/errors/38d1ad697be3d16f. Report an issue: GitHub.