Fix eval_imagenet and add eval_winoground

[ajtejankar]

1. Replace .cuda() calls with .to() calls and correct device in eval_imagenet function
2. Add eval_winoground to evaluate on the Winoground dataset

[anas-awadalla]

Thanks for the contribution @ajtejankar! I assigned @jpgard, our expert on classification tasks :).

[ajtejankar]

It seems the linter is failing on multiple pull requests not just mine. Do you know if it is the linter or I need to fix my pull request?

[jpgard]

Thanks for the PR @ajtejankar !

I am going to leave this for now -- we are in the middle of a large refactor of our classification code to support more tasks and fix some other bugs. Once https://github.com/mlfoundations/open_flamingo/tree/text-caching goes in, I will revisit this one.

[jpgard]

(yes, you probably need to lint your code, our linter is strict and complains a lot -- we use the black linter.)

[anas-awadalla]

(yes, you probably need to lint your code, our linter is strict and complains a lot -- we use the black linter.)

Can you make sure you have the latest version of black installed and then run "black ."? That should fix all formatting issues.

[ajtejankar]

Sure, we can integrate this later. Also, thanks for the help with linter. For now, I will update the pull request after running black.