OllamaLocal LLM
Larger LLMs Confidently Hallucinate: 120B Test Breaks 'Bigger Is Better' Myth
Reddit tests 1.5B–120B models: large ones output identical wrong answers when hallucinating. 'Bigger is better' myth broken for critical AI tasks.
Oct 2·2 min read
