ValueError when using swin_upernet models

[SUC-DriverOld]

ValueError when using swin_upernet models, The solution can refer to ZFTurbo/Music-Source-Separation-Training/issues/6

ValueError: Make sure that the channel dimension of the pixel values match with the one set in the configuration.

[SUC-DriverOld]

ValueError when using swin_upernet models

ValueError: Make sure that the channel dimension of the pixel values match with the one set in the configuration.

Follow the steps below to solve this problem:

1. Go to site-packages\transformers\models\swin\modeling_swin.py
2. Modify the function code on line 312 to the following：

    def forward(self, pixel_values: Optional[torch.FloatTensor]) -> Tuple[torch.Tensor, Tuple[int]]:
        _, num_channels, height, width = pixel_values.shape
        if num_channels != self.num_channels:
            self.num_channels = num_channels
        # pad the input to be divisible by self.patch_size, if needed
        pixel_values = self.maybe_pad(pixel_values, height, width)
        embeddings = self.projection(pixel_values)
        _, _, height, width = embeddings.shape
        output_dimensions = (height, width)
        embeddings = embeddings.flatten(2).transpose(1, 2)

        return embeddings, output_dimensions