keras-team/keras · error · ValueError

`sparse` may only be true if `output_mode` is `'one_hot'`, `

Error message

`sparse` may only be true if `output_mode` is `'one_hot'`, `'multi_hot'`, `'count'` or `'tf_idf'`. Received: sparse={sparse} and output_mode={output_mode}

What it means

Integer output mode produces a dense tensor of indices, so `sparse=True` is meaningless there. Sparse output is only supported for the wide modes: one_hot, multi_hot, count and tf_idf.

Source

Thrown at keras/src/layers/preprocessing/index_lookup.py:202

            allowable_strings=(
                "int",
                "one_hot",
                "multi_hot",
                "count",
                "tf_idf",
            ),
            caller_name=self.__class__.__name__,
            arg_name="output_mode",
        )

        if invert and output_mode != "int":
            raise ValueError(
                "`output_mode` must be `'int'` when `invert` is true. "
                f"Received: output_mode={output_mode}"
            )

        if sparse and output_mode == "int":
            raise ValueError(
                "`sparse` may only be true if `output_mode` is "
                "`'one_hot'`, `'multi_hot'`, `'count'` or `'tf_idf'`. "
                f"Received: sparse={sparse} and "
                f"output_mode={output_mode}"
            )

        if idf_weights is not None and output_mode != "tf_idf":
            raise ValueError(
                "`idf_weights` should only be set if `output_mode` is "
                f"`'tf_idf'`. Received: idf_weights={idf_weights} and "
                f"output_mode={output_mode}"
            )

        super().__init__(name=name)
        self._convert_input_args = False
        self._allow_non_tensor_positional_args = True
        self.supports_jit = False

View on GitHub (pinned to 7a34a03db6)

Solutions

1. Drop `sparse=True` when you need integer indices.
2. Or switch output_mode to 'multi_hot', 'count' or 'tf_idf' to get sparse output.
3. For memory-efficient pipelines, keep int output and let the Embedding layer handle large vocabs.

Example fix

# before
layer = IndexLookup(sparse=True)  # output_mode defaults to 'int'

# after
layer = IndexLookup(output_mode='multi_hot', sparse=True)

if sparse and output_mode == 'int':
    raise ValueError('sparse requires one_hot/multi_hot/count/tf_idf output')
layer = IndexLookup(output_mode=output_mode, sparse=sparse)

def sparse_mode_ok(mode) -> bool:
    return mode in ('one_hot', 'multi_hot', 'count', 'tf_idf')

When it happens

Trigger: `IndexLookup(sparse=True, output_mode='int')` — including the default output_mode='int' when only `sparse` is set.

Common situations: Enabling sparse output to save memory on large vocabularies while leaving output_mode at its default.

Related errors

• `num_oov_indices` must be greater than or equal to 0. Receiv
• `salt` can only be used when `oov_method='farmhash'`. Receiv
• The `salt` argument for `IndexLookup` can only be a tuple of
• `output_mode` must be `'int'` when `invert` is true. Receive
• `idf_weights` should only be set if `output_mode` is `'tf_id

AI-assisted analysis of keras-team/keras@7a34a03db6 (2026-08-25). Data as JSON: /api/errors/1e892db3803582e8. Report an issue: GitHub.