Post

What happened to those retro frontier models only trained on text pre-1900s? Let’s see if they can make discoveries in chemistry and discover quantum physics as a possibility. Where do they miss? Are there alternative tracks to the tech tree that we missed? x.com/emollick/st...

27/10/26, 3:05 pm0 reposts1 like

Daniel Mewes

@dmewes.com

I imagine their scale (both data and model size) might be too small for that, plus you'd need to figure out how to post-train them in a century-appropriate way (call file editing tools and run Python code doesn't really work for obvious reasons - maybe some other domains could?).