hardcoding metrics (for now)

Initial version of the Evaluation Buddy -- currently most things are hardcoded (e.g. the dataset list), but the goal it to make it all compatible with the Hub!