I think there’s room for private tests, but those should probably be done by some kind of independent standards body. LLMs are an incredible advancement of mach

I think there’s room for private tests, but those should probably be done by some kind of independent standards body. LLMs are an incredible advancement of machine learning research, but we diminish that when we let these kind of wholly unscientific evaluations predominate and especially in open source, where so much great work is otherwise being done to combat the monopoly tactics of big tech.