There is no new idea between a tiny model and the biggest one on earth. Same loop, same
training, just far more text and far more internal settings. What changes is what starts
working without anyone asking for it. People call that emergence.
The same design, made enormous. Nobody added the new abilities on the big one; past a certain size they simply turned up.
1Pick a model size
2Pick something to ask it
Do thisLeave the question the same and click up through the four sizes
in step 1. Nothing about the method changes, only how much of it there is.
