Is the MMI Criterion Necessary for Interpretability? Degenerating Non-causal Features to Plain Noise for Self-Rationalization

Documenti analoghi

• Breaking Free from MMI: A New Frontier in Rationalization by Probing Input Utilization
  di: Liu, Wei, et al.
  Pubblicazione: (2025)
• Adversarial Cooperative Rationalization: The Risk of Spurious Correlations in Even Clean Datasets
  di: Liu, Wei, et al.
  Pubblicazione: (2025)
• Adversarial Attack for Explanation Robustness of Rationalization Models
  di: Zhang, Yuankai, et al.
  Pubblicazione: (2024)
• Enhancing RAG with Active Learning on Conversation Records: Reject Incapables and Answer Capables
  di: Geng, Xuzhao, et al.
  Pubblicazione: (2025)
• Is Model Editing Built on Sand? Revealing Its Illusory Success and Fragile Foundation
  di: Liu, Wei, et al.
  Pubblicazione: (2025)