I think you're right. In general, storage and operating formats seem to be decoupling for AI/ML. Nvidia's E8M12 is also a format specifically for operators - t

I think you're right. In general, storage and operating formats seem to be decoupling for AI/ML. Nvidia's E8M12 is also a format specifically for operators - they expect you to store FP32 when you operate in E8M12. Storage is almost always in power-of-2 sizes.