keras-team/keras · error · ValueError

Invalid value for argument `output_mode`. Expected one of {a

Error message

Invalid value for argument `output_mode`. Expected one of {accepted_output_modes}. Received: output_mode={output_mode}

What it means

The Hashing preprocessing layer only supports four output modes: 'int', 'one_hot', 'multi_hot', and 'count'. The layer validates output_mode in __init__ and raises this ValueError when any other string (or a typo) is passed. Internally the mode decides how the hashed bucket index is converted into an output tensor.

Source

Thrown at keras/src/layers/preprocessing/hashing.py:177

        if num_bins is None or num_bins <= 0:
            raise ValueError(
                "The `num_bins` for `Hashing` cannot be `None` or "
                f"non-positive values. Received: num_bins={num_bins}."
            )

        if output_mode == "int" and (
            self.dtype_policy.name not in ("int32", "int64")
        ):
            raise ValueError(
                'When `output_mode="int"`, `dtype` should be an integer '
                f"type, 'int32' or 'in64'. Received: dtype={kwargs['dtype']}"
            )

        # 'output_mode' must be one of (INT, ONE_HOT, MULTI_HOT, COUNT)
        accepted_output_modes = ("int", "one_hot", "multi_hot", "count")
        if output_mode not in accepted_output_modes:
            raise ValueError(
                "Invalid value for argument `output_mode`. "
                f"Expected one of {accepted_output_modes}. "
                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"`, or `"count"`. '
                f"Received: sparse={sparse} and "
                f"output_mode={output_mode}"
            )

        self.num_bins = num_bins
        self.mask_value = mask_value
        self.strong_hash = True if salt is not None else False
        self.output_mode = output_mode
        self.sparse = sparse

View on GitHub (pinned to 7a34a03db6)

Solutions

Set output_mode to one of exactly 'int', 'one_hot', 'multi_hot', or 'count'.
Note that 'tf_idf' is not supported by Hashing; use TextVectorization if you need tf-idf output.
Check for leading/trailing whitespace or wrong case in the string you pass programmatically.

Example fix

# before
layer = keras.layers.Hashing(num_bins=100, output_mode="onehot")
# after
layer = keras.layers.Hashing(num_bins=100, output_mode="one_hot")

Defensive patterns

Strategy: validation

Validate before calling

ALLOWED = {"int", "one_hot", "multi_hot", "count"}
if output_mode not in ALLOWED:
    raise ValueError(f"unsupported output_mode {output_mode!r}")

Type guard

def is_valid_output_mode(m):
    return isinstance(m, str) and m in {"int", "one_hot", "multi_hot", "count"}

Prevention

Keep output mode names in a module-level constant set and reuse it at every call site.
Add a unit test that constructs the layer with each supported mode.

When it happens

Trigger: Calling keras.layers.Hashing(output_mode=...) with a misspelled or unsupported value, e.g. 'onehot', 'multi-hot', 'INT', 'freq', 'binary', or passing None.

Common situations: Porting code from tf.keras StringLookup/IntegerLookup (which accept 'tf_idf' and other modes) to the standalone Hashing layer; typos from camelCase vs snake_case mode names.

Related errors

AI-assisted analysis of keras-team/keras@7a34a03db6 (2026-08-25). Data as JSON: /api/errors/32036690996bbaf8. Report an issue: GitHub.