Tbh a lot of ai conference papers would absolutely be writable from most engineers. A lot of papers are : we trained a model and got these results. They didn't publish the weight though, so you would be unable to check the paper.