henrique

joined 1 year ago
[–] henrique@lemmy.world 10 points 1 year ago

Made an integration with one LLM recently and had the same thought. Too many inputs and the responses are incorrect. Too few and it starts to hallucinate.