@LocBibliophilia @BetleyJan @ASM65617010 @OwainEvans_UK @cloud_kx @minhxle1 @jameschua_sg @anna_sztyber @saprmarks Or even more pretraining. I expect this would still work.
For instance, I think Sonnet 3.5-4 are likely the same base model in this sense but updated with more pretraining and more/different posttraining. Opus 3 & 4 might be too.
For instance, I think Sonnet 3.5-4 are likely the same base model in this sense but updated with more pretraining and more/different posttraining. Opus 3 & 4 might be too.