Using a CNN on a tabular dataloader

I am playing around with Fastai on the Colab MNIST sample data. The data is already in tabular form, with each column representing a pixel, so I’m pretty sure I cannot use an image loader. I loaded it into a tabular dataloader and trained it to a 95% accuracy, but I would like to be able to use a CNN. I tried just creating a cnn learner with the tabular dls, but I am getting this: TypeError: forward() takes 2 positional arguments but 3 were given. What is the easiest way to use a CNN on a tabular dataloader? Or can I convert the tabular dataloader to an image dataloader? Or am I doing it correctly and this error is because of the data I’m using?