[MuffinFlavored]

> - Finetuned local LLMs are good for specific niches, like roleplaying, text games, and helper bots for your own pile of data.

I can't see how they don't hallucinate/are leagues away from GPT-3.5 let alone GPT-4 level of quality of output. Am I mistaken?

[brucethemoose2]

They are better than GPT 3.5 (which I am generally not impressed with), but not as good as GPT4.

Again, the specialized variants perform very well in their niches.

[astrange]

Hallucinations are exactly what you want in a gaming model. That's another way of saying "creativity".

[gmerc]

You seem to assume hallucinations are a fatal flaw. You give it a document to summarize and see how often it hallucinates. Very little. Human performance.

Now often does a human make random shit up about general knowledge questions?