I propose we move away from the framing of "Model years" - they're standard human research years. Yes, likely more people are working on it, and also working ha

I propose we move away from the framing of "Model years" - they're standard human research years. Yes, likely more people are working on it, and also working harder, but ever since we acquired a certain amount of compute in the world, many people were able to independently find the same patterns and train models.