Hallucinations in Large Language Models occur when the probabilistic model incorrectly favors a hallucinatory output (yhalluc) over a factually correct response (yfact), representing a mismatch between the model's internal probability distributions and real-world factual distributions.

Authors

• Person: Not available Organization: Frontiers
  Survey and analysis of hallucinations in large language models

Sources

• Survey and analysis of hallucinations in large language models www.frontiersin.org Frontiers via serper

Referenced by nodes (2)

• Large Language Models concept
• hallucination concept