Processor of Natural Languages · Jun 12

I like "we tried simple approach and it works well, turns out we were overcomplicating things" papers usually, but its this every single time. You spin mundane things like "model diffing" (finding a difference between two models) into some existential problem, then this can only be the consequence.