I never tracked is like news. I used it for two main things: - Checking the state of the art (SotA) for a given problem. For some problems 2 year old solutions

I never tracked is like news. I used it for two main things: - Checking the state of the art (SotA) for a given problem. For some problems 2 year old solutions are still close to SotA; in others - there is a huge difference. And if there is a huge difference - is it because of architecture and parameter tuning, or using totally different architectures and training modes. - Running code - to be used somewhere, or as a reference. Papers never have all details, and do not compile. Context: I used to work in the field, as a consultant. Though, I cite Papers with Code in one overview paper.