A study found that the Trustworthy Language Model (TLM) detects incorrect responses more effectively than LLM-as-a-judge or token probability (logprobs) techniques across all major LLM models.

Authors

• Person: Not available Organization: Cleanlab
  Benchmarking Hallucination Detection Methods in RAG - Cleanlab

Sources

• Benchmarking Hallucination Detection Methods in RAG - Cleanlab cleanlab.ai Cleanlab via serper

Referenced by nodes (2)

• LLM-as-a-judge concept
• Trustworthy Language Model concept