Why do LLaVA Vision-Language Models Reply to Images in English?

Documenti analoghi

• LLaVA-Gemma: Accelerating Multimodal Foundation Models with a Compact Language Model
  di: Hinck, Musashi, et al.
  Pubblicazione: (2024)
• ClimDetect: A Benchmark Dataset for Climate Change Detection and Attribution
  di: Yu, Sungduk, et al.
  Pubblicazione: (2024)
• Debiasing Large Vision-Language Models by Ablating Protected Attribute Representations
  di: Ratzlaff, Neale, et al.
  Pubblicazione: (2024)
• Steering Large Language Models to Evaluate and Amplify Creativity
  di: Olson, Matthew Lyle, et al.
  Pubblicazione: (2024)
• Probing Semantic Routing in Large Mixture-of-Expert Models
  di: Olson, Matthew Lyle, et al.
  Pubblicazione: (2025)