Re:NanoGPT Slowrun: 10x Data Efficiency with Infinite Compute
by ColonelPhantom on Friday March 20, 2026 @09:28AM
(#47454206)
Interesting; I was not aware of those "universal synthetics" but they make sense: a stronger reasoning base would make modeling tasks easier. Thanks for the link!
Again, though, if those work I assume they will be used for the slowrun. Surely a few hundred LoC to generate data would not be considered cheating :)
Just the other day this was posted, for example: https://news.ycombinator.com/item?id=47388293