Large Vision-Language Model (LVLM) hallucinations originate from three interacting causal pathways: image-to-input-text, image-to-output-text, and text-to-text.

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 (2)

• hallucination concept
• Large Vision-Language Models concept