I'd be shocked if we don't see diminishing returns in the inference compute scaling laws. We already didn't deserve how clean and predictive the pre-training sc

I'd be shocked if we don't see diminishing returns in the inference compute scaling laws. We already didn't deserve how clean and predictive the pre-training scaling laws were, no way the universe grants us another boon of that magnitude