[fao_]

Even with reinforcement learning, you can still find phrases and patterns that are repeated in the smaller models. It's likely true with the larger ones, too, except the corpus is so large that you'll have fat luck to pick out which specific bits.

[RugnirViking]

what?

what do you mean? are you claiming its hard to recognize the features of speech of large models? its really not. There are famous wikipedia articles about it. Heck an em dash, a single character is often a pretty good clue

[fao_]

That is not what I was saying at all :P