Literally Anything

The model scores completed maps using known roles and lanes, without results or duration as inputs. Some maps may have been used for training; this is not an independent evaluation.