Beg to Differ: Understanding Reasoning-Answer Misalignment Across Languages

Documenti analoghi

• The Root Shapes the Fruit: On the Persistence of Gender-Exclusive Harms in Aligned Language Models
  di: Ovalle, Anaelia, et al.
  Pubblicazione: (2024)
• Changing Answer Order Can Decrease MMLU Accuracy
  di: Gupta, Vipul, et al.
  Pubblicazione: (2024)
• What's in Common? Multimodal Models Hallucinate When Reasoning Across Scenes
  di: Ross, Candace, et al.
  Pubblicazione: (2025)
• Chained Tuning Leads to Biased Forgetting
  di: Ung, Megan, et al.
  Pubblicazione: (2024)
• What makes a good metric? Evaluating automatic metrics for text-to-image consistency
  di: Ross, Candace, et al.
  Pubblicazione: (2024)