keras-team/keras · error · ValueError

All of the arrays in `x` should have the same length. Found

Error message

All of the arrays in `x` should have the same length. Found a pair with: len(x[0]) = {len(x)}, len(x[?]) = {len(xx)}

What it means

NumpyArrayIterator.__init__ accepts x as a tuple of (images, [aux arrays...]); every auxiliary array must have the same first-dimension length as the images. A mismatch raises this ValueError because batching would produce misaligned multi-input batches.

Source

Thrown at keras/src/legacy/preprocessing/image.py:556

        save_format="png",
        subset=None,
        ignore_class_split=False,
        dtype=None,
    ):
        if data_format is None:
            data_format = backend.image_data_format()
        if dtype is None:
            dtype = backend.floatx()
        self.dtype = dtype
        if isinstance(x, tuple) or isinstance(x, list):
            if not isinstance(x[1], list):
                x_misc = [np.asarray(x[1])]
            else:
                x_misc = [np.asarray(xx) for xx in x[1]]
            x = x[0]
            for xx in x_misc:
                if len(x) != len(xx):
                    raise ValueError(
                        "All of the arrays in `x` "
                        "should have the same length. "
                        "Found a pair with: "
                        f"len(x[0]) = {len(x)}, len(x[?]) = {len(xx)}"
                    )
        else:
            x_misc = []

        if y is not None and len(x) != len(y):
            raise ValueError(
                "`x` (images tensor) and `y` (labels) "
                "should have the same length. "
                f"Found: x.shape = {np.asarray(x).shape}, "
                f"y.shape = {np.asarray(y).shape}"
            )
        if sample_weight is not None and len(x) != len(sample_weight):
            raise ValueError(
                "`x` (images tensor) and `sample_weight` "

View on GitHub (pinned to 7a34a03db6)

Solutions

1. Assert lengths match before calling: all(len(a) == len(x_imgs) for a in aux)
2. Re-derive all inputs from the same index/mask so they stay aligned
3. Shuffle with a shared permutation applied to every array

Example fix

# before
it = gen.flow((x_imgs, meta), y)  # len(meta) != len(x_imgs)

# after
assert len(x_imgs) == len(meta) == len(y)
it = gen.flow((x_imgs, meta), y)

n = len(x[0])
assert all(len(a) == n for a in x[1]), [len(a) for a in x[1]]

def aligned_inputs(x):
    n = len(x[0])
    return all(len(a) == n for a in x[1])

When it happens

Trigger: flow(x=(x_imgs, meta_array), y=y) where meta_array has fewer or more rows than x_imgs; slicing or filtering one input but not the other after a train/test split.

Common situations: Multi-input models (image + tabular metadata) where one array was shuffled, subsampled, or filtered independently; off-by-one errors after dropping NaN rows from only part of the inputs.

Related errors

• `x` (images tensor) and `y` (labels) should have the same le
• Architecture configuration does not match {weights_name} var
• Model name "{name}" does not match weights variant "{weights
• DenseNet does not support the `channels_first` image data fo
• The last dimension of `query_shape` and `value_shape` must b

AI-assisted analysis of keras-team/keras@7a34a03db6 (2026-08-25). Data as JSON: /api/errors/5dbdaaec105a1a17. Report an issue: GitHub.