Post #1753541
2026-04-19 06:54 UTC
@astraleureka The analogy is messy because the tokens the model emits aren't the same thing as the character strings they're converted to/from. Like the tag for initiating chain-of-thought is a single token that exists for that purpose, and is not the same as the multiple tokens that would spell it out character-by-character. That token is out-of-band in the same way as the comma and control symbols in 8b10b are out-of-band. Tool calls are the same.
The other problem you pointed out is probably the bigger one which is we took Turing machines, made the tape append-only, added associative lookups on the tape and poured the entire internet into them until they have anxiety, and the fact they appear to follow natural-language instructions most of the time is a coincidence. Having out-of-band control symbols is nice but there's no way to actually know or control when they're emitted.
Replies (1)
-
@crzwdjk@mastodon.social 2026-04-19 17:13
@wren6991 @astraleureka I think the problem is not that there are no out of band control signals, it's that the response is ultimately based on input text and there's no way to reliably separate user instructions from other text.