Large language models can do something very very small language models have done since forever.

I don’t feel like shitting on this one too hard since I guess it’s a mildly interesting variation on a Markov chain LLM, but the title felt extremely sneerworthy.

I’m giving them the benefit of the doubt because their README is too tiring to read for me to figure out what this might be used for. That’s coming from someone who spent most of today reading SPARC assembly for fun.

Embarrassed myself by accidentally posting this to some other instance somehow. Stupid janky Lemmy offering communities I've never even looked at right in the posting interface.