keras-team/keras · error · ValueError

Input should have rank >= 1. Received: input.shape = {real.s

Error message

Input should have rank >= 1. Received: input.shape = {real.shape}

What it means

keras.ops.fft computes a 1D FFT over the last axis, so every input (real and imaginary part) must have at least one dimension. A rank-0 scalar tensor fails this check in compute_output_spec.

Source

Thrown at keras/src/ops/math.py:518

        if not isinstance(x, (tuple, list)) or len(x) != 2:
            raise ValueError(
                "Input `x` should be a tuple of two tensors - real and "
                f"imaginary. Received: x={x}"
            )

        real, imag = x
        # Both real and imaginary parts should have the same shape.
        if real.shape != imag.shape:
            raise ValueError(
                "Input `x` should be a tuple of two tensors - real and "
                "imaginary. Both the real and imaginary parts should have the "
                f"same shape. Received: x[0].shape = {real.shape}, "
                f"x[1].shape = {imag.shape}"
            )

        # We are calculating 1D FFT. Hence, rank >= 1.
        if len(real.shape) < 1:
            raise ValueError(
                f"Input should have rank >= 1. "
                f"Received: input.shape = {real.shape}"
            )

        # The axis along which we are calculating FFT should be fully-defined.
        m = real.shape[-1]
        if m is None:
            raise ValueError(
                f"Input should have its last dimension fully-defined. "
                f"Received: input.shape = {real.shape}"
            )

        return (
            KerasTensor(shape=real.shape, dtype=real.dtype),
            KerasTensor(shape=imag.shape, dtype=imag.dtype),
        )

    def call(self, x):

View on GitHub (pinned to 7a34a03db6)

Solutions

1. Keep at least a length-1 vector: reshape scalars to shape (1,) or (n,) before the call
2. Audit reductions upstream (mean/sum/max) and add keepdims=True
3. Use keras.ops.expand_dims(x, -1) if the signal may be scalar

Example fix

# before
sig = keras.ops.mean(x)  # scalar
out = keras.ops.fft((sig, keras.ops.zeros_like(sig)))
# after
sig = keras.ops.mean(x, keepdims=True)  # shape (1,)
out = keras.ops.fft((sig, keras.ops.zeros_like(sig)))

assert len(real.shape) >= 1 and len(imag.shape) >= 1

When it happens

Trigger: Passing scalar tensors like keras.ops.cast(3.0, 'float32') as either part; aggressive reduction (sum/mean without keepdims) collapsing the signal to a scalar before the FFT.

Common situations: Preprocessing that aggregates the time series (mean/variance normalization) forgetting keepdims=True, or unit tests using trivial scalar placeholders.

Related errors

• Input `x` should be a tuple of two tensors - real and imagin
• Input should have rank >= 2. Received: input.shape = {real.s
• Layer {self.name} weight shape {variable.shape} is not compa
• Expected rebatched data to have batch size 1. Received: shap
• Expected as input a list/tuple of 2 tensors. Received input_

AI-assisted analysis of keras-team/keras@7a34a03db6 (2026-08-25). Data as JSON: /api/errors/04ebfaeead893c0a. Report an issue: GitHub.