[shironandonon_]

Only 29% is pretty good I would say. I personally would much-prefer to run a dumber local LLM that hallucinates a bit and isn’t too woke.

[cypress66]

Then use llama?