keras-team/keras · error · ValueError

`slice_update` requires integral start_indices

Error message

`slice_update` requires integral start_indices

What it means

Error "`slice_update` requires integral start_indices" thrown in keras-team/keras.

Source

Thrown at keras/src/backend/openvino/core.py:1445

    if isinstance(start_indices, (list, np.ndarray)):
        start_indices = tuple(start_indices)
    if not isinstance(start_indices, tuple):
        raise ValueError(
            "`slice_update` is not supported by openvino backend"
            " for `start_indices` of type {}".format(type(start_indices))
        )

    zero_scalar = ov_opset.constant(0, Type.i32)
    one_scalar = ov_opset.constant(1, Type.i32)
    zero_tensor = ov_opset.constant([0], Type.i32)
    one_tensor = ov_opset.constant([1], Type.i32)

    processed_start_indices = []
    for idx in start_indices:
        val = get_ov_output(idx)
        if not val.get_element_type().is_integral():
            raise ValueError("`slice_update` requires integral start_indices")
        if val.get_element_type() != Type.i32:
            val = ov_opset.convert(val, Type.i32).output(0)
        if val.get_partial_shape().rank.get_length() == 0:
            val = ov_opset.unsqueeze(val, zero_scalar).output(0)
        processed_start_indices.append(val)

    updates_shape = ov_opset.shape_of(updates_tensor, Type.i32).output(0)
    rank = updates_tensor.get_partial_shape().rank.get_length()
    if rank == 0:
        # Handle scalar update
        start_tensor = ov_opset.concat(processed_start_indices, axis=0).output(
            0
        )
        # For scatter_nd_update,
        # indices should be of shape [num_updates, rank_of_inputs]
        # and updates should be of shape [num_updates]. Here num_updates is 1.
        absolute_indices = ov_opset.unsqueeze(start_tensor, zero_scalar).output(
            0

View on GitHub (pinned to 7a34a03db6)

When it happens

Trigger: Thrown at keras/src/backend/openvino/core.py:1445 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/d57a152731a233fe. Report an issue: GitHub.