keras-team/keras · error · ValueError

All `HashedCrossing` inputs should have shape `()`, `(batch_

Error message

All `HashedCrossing` inputs should have shape `()`, `(batch_size)` or `(batch_size, 1)`. Received: inputs={inputs}

What it means

HashedCrossing only accepts scalar per-sample inputs: shape (), (batch_size,) or (batch_size, 1). Tensors with rank > 2, or rank 2 with last dimension != 1, are rejected because each sample must contribute exactly one value to the crossing.

Source

Thrown at keras/src/layers/preprocessing/hashed_crossing.py:205

        }

    def _check_at_least_two_inputs(self, inputs):
        if not isinstance(inputs, (list, tuple)):
            raise ValueError(
                "`HashedCrossing` should be called on a list or tuple of "
                f"inputs. Received: inputs={inputs}"
            )
        if len(inputs) < 2:
            raise ValueError(
                "`HashedCrossing` should be called on at least two inputs. "
                f"Received: inputs={inputs}"
            )

    def _check_input_shape_and_type(self, inputs):
        first_shape = tuple(inputs[0].shape)
        rank = len(first_shape)
        if rank > 2 or (rank == 2 and first_shape[-1] != 1):
            raise ValueError(
                "All `HashedCrossing` inputs should have shape `()`, "
                "`(batch_size)` or `(batch_size, 1)`. "
                f"Received: inputs={inputs}"
            )
        if not all(tuple(x.shape) == first_shape for x in inputs[1:]):
            raise ValueError(
                "All `HashedCrossing` inputs should have equal shape. "
                f"Received: inputs={inputs}"
            )
        if any(
            isinstance(x, (tf.RaggedTensor, tf.SparseTensor)) for x in inputs
        ):
            raise ValueError(
                "All `HashedCrossing` inputs should be dense tensors. "
                f"Received: inputs={inputs}"
            )
        if not all(
            tf.as_dtype(x.dtype).is_integer or x.dtype == tf.string

View on GitHub (pinned to 7a34a03db6)

Solutions

1. Reshape each input to (batch,) or (batch, 1): tf.reshape(x, [-1]) or [-1, 1]
2. Reduce multi-value features to a single value before crossing (e.g. take the first token or aggregate)
3. Check upstream layers (Embedding, dense features) are not producing (batch, k>1)

Example fix

// before
out = layer([a, b])  # a.shape == (None, 3)
// after
a = keras.layers.Reshape((1,))(a)  # or select/slice to one column first
out = layer([a, b])

for x in inputs:
    assert len(x.shape) <= 1 or (len(x.shape) == 2 and int(x.shape[-1]) == 1)

def acceptable_shape(t):
    s = tuple(t.shape)
    return len(s) <= 1 or (len(s) == 2 and s[-1] == 1)

catch ValueError from call() and reshape each input to (batch, 1) with tf.reshape(x, [-1, 1]) before retrying

When it happens

Trigger: An input with rank > 2, or rank 2 whose last dimension != 1 (e.g. shape (batch, 5)) reaching _check_input_shape_and_type.

Common situations: Feeding multi-column features shape (batch, n>1); feeding 3D tensors from sequence pipelines; forgetting to slice a wide DataFrame column set down to one value.

Related errors

• Expected as input a list/tuple of 2 tensors. Received input_
• Expected the two input tensors to have identical shapes. Rec
• All `HashedCrossing` inputs should have equal shape. Receive
• Layer {self.name} weight shape {variable.shape} is not compa
• Expected rebatched data to have batch size 1. Received: shap

AI-assisted analysis of keras-team/keras@7a34a03db6 (2026-08-25). Data as JSON: /api/errors/0352234ff7c917b1. Report an issue: GitHub.