The BAFH framework is a lightweight method that trains a feedforward classifier on hidden states of Large Language Models to determine belief states and classify hallucination types, as evaluated against MIND and SAR baselines using Gemma-2, Llama-3.1, and Mistral models.

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 (4)

• Large Language Models concept
• LLaMA concept
• mind concept
• Mistral AI entity