Interventions targeting factuality in AI models often degrade faithfulness, and interventions targeting faithfulness often degrade factuality, creating a zero-sum dynamic.

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 (1)

• AI models concept