GPT-4 achieves an F1-score of approximately 0.625 in detecting subtle falsehoods on the hardest subset of the MedHallu benchmark.

Authors

• Person: Not available Organization: GitHub
  EdinburghNLP/awesome-hallucination-detection - GitHub

Sources

• EdinburghNLP/awesome-hallucination-detection - GitHub github.com GitHub via serper

Referenced by nodes (3)

• GPT-4 concept
• MedHallu concept
• F1 score concept