huggingface/transformers · error · ValueError

Size must contain 'height' and 'width' keys, or 'max_height'

Error message

Size must contain 'height' and 'width' keys, or 'max_height' and 'max_width', or 'shortest_edge' key. Got {size}.

What it means

Error "Size must contain 'height' and 'width' keys, or 'max_height' and 'max_width', or 'shortest_edge' key. Got {size}." thrown in huggingface/transformers.

Source

Thrown at src/transformers/image_processing_backends.py:249

        if size.shortest_edge and size.longest_edge:
            new_size = get_size_with_aspect_ratio(
                image.size()[-2:],
                size.shortest_edge,
                size.longest_edge,
            )
        elif size.shortest_edge:
            new_size = get_resize_output_image_size(
                image,
                size=size.shortest_edge,
                default_to_square=False,
                input_data_format=ChannelDimension.FIRST,
            )
        elif size.max_height and size.max_width:
            new_size = get_image_size_for_max_height_width(image.size()[-2:], size.max_height, size.max_width)
        elif size.height and size.width:
            new_size = (size.height, size.width)
        else:
            raise ValueError(
                "Size must contain 'height' and 'width' keys, or 'max_height' and 'max_width', or 'shortest_edge' key. Got"
                f" {size}."
            )

        # Workaround for torch.compile issue with uint8 on AMD GPUs
        if is_torchdynamo_compiling() and is_rocm_platform():
            return self._compile_friendly_resize(image, new_size, interpolation, antialias)
        return tvF.resize(image, new_size, interpolation=interpolation, antialias=antialias)

    @staticmethod
    def _compile_friendly_resize(
        image: "torch.Tensor",
        new_size: tuple[int, int],
        interpolation: Optional["tvF.InterpolationMode"] = None,
        antialias: bool = True,
    ) -> "torch.Tensor":
        """A wrapper around tvF.resize for torch.compile compatibility with uint8 tensors."""
        if image.dtype == torch.uint8:

View on GitHub (pinned to a597f97485)

Solutions

1. Provide size with 'height'/'width', 'max_height'/'max_width', or 'shortest_edge' keys.
2. Use `get_size_dict` to normalize the size argument.

When it happens

Trigger: Raised in image resize preprocessing when the size dict lacks a recognized key combination.

Common situations: size dict without ('height','width'), ('max_height','max_width'), or 'shortest_edge' keys passed to a resize backend.

AI-assisted analysis of huggingface/transformers@a597f97485 (2026-08-14). Data as JSON: /api/errors/9f838699bfb105a5. Report an issue: GitHub.