Grounded Generation, Hallucinations & LLM Evaluation: Assessing Model Output Quality
Video by H2O.ai via YouTube AI can be confidently wrong. Here is why it happens, how grounding fixes it, and how to measure whether it worked. Terms covered: – Hallucination: models are trained to generate statistically likely text, not to verify facts. When a model does not know, it fills the gap with something plausible. … Read more