[mncharity]

> rerunning [...] but differently framed or worded input, and seeing how much they diverged

I'm surprised how little attention this is getting in today's comments. Open-weights means being able to afford multiple runs, space sampling, critique and synthesis.

Last night, sketching some intro biology content emphasizing cross-cutting concerns, Qwen would get sucked into the Next Generation Science Standards attractor. But nudge it by adding just one similar phrase of Chinese, and most runs ran free (outputting English but for headers with parenthesized Chinese). The multi-lingual LLM "no, not *that* region of latent space".