Can a Classifier Chat? Turning Jev into a (Tiny) LLM
A fun experiment: can Jev, a classifier that picks from a fixed set of choices, behave like a traditional LLM and hold a conversation? I tried one letter at a time, one word at a time, and finally a 1,700-word vocabulary split into categories. It's clearly not what Jev is built for, but here's what worked, what didn't, and why.



