keras-team/keras · error · ValueError

Cannot use array_split with static Python logic on dynamic a

Error message

Cannot use array_split with static Python logic on dynamic axis. Axis {axis} has unknown dimension for shape {original_shape}.

What it means

Error "Cannot use array_split with static Python logic on dynamic axis. Axis {axis} has unknown dimension for shape {original_shape}." thrown in keras-team/keras.

Source

Thrown at keras/src/backend/openvino/numpy.py:4398

        splits = ov_opset.variadic_split(x, axis_tensor, length_tensor)
        result = []
        for i in range(len(split_lengths) + 1):
            result.append(OpenVINOKerasTensor(splits.output(i)))
        return result

    raise TypeError(
        f"unsupported type of indices_or_sections: {type(indices_or_sections)}"
    )


def array_split(x, indices_or_sections, axis=0):
    original_shape = x.shape
    x = get_ov_output(x)

    num_splits_val = indices_or_sections
    total_size = original_shape[axis]
    if total_size is None:
        raise ValueError(
            f"Cannot use array_split with static Python logic on dynamic axis. "
            f"Axis {axis} has unknown dimension for shape {original_shape}."
        )

    base_size = total_size // num_splits_val
    remainder = total_size % num_splits_val

    split_lengths = [base_size + 1] * remainder + [base_size] * (
        num_splits_val - remainder
    )
    split_lengths_tensor = ov_opset.constant(
        split_lengths, dtype=Type.i64
    ).output(0)

    axis_tensor = ov_opset.constant(axis, dtype=Type.i32).output(0)
    splits = ov_opset.variadic_split(x, axis_tensor, split_lengths_tensor)

    result = []

View on GitHub (pinned to 7a34a03db6)

When it happens

Trigger: Thrown at keras/src/backend/openvino/numpy.py:4398 when the library encounters an invalid state.

Common situations: See trigger scenarios.

AI-assisted analysis of keras-team/keras@7a34a03db6 (2026-08-25). Data as JSON: /api/errors/5e5637452aa32190. Report an issue: GitHub.