Reasoning Bias of Next Token Prediction Training

Documenti analoghi

• Complexity Control Facilitates Reasoning-Based Compositional Generalization in Transformers
  di: Zhang, Zhongwang, et al.
  Pubblicazione: (2025)
• An Analysis for Reasoning Bias of Language Models with Small Initialization
  di: Yao, Junjie, et al.
  Pubblicazione: (2025)
• Initialization is Critical to Whether Transformers Fit Composite Functions by Reasoning or Memorizing
  di: Zhang, Zhongwang, et al.
  Pubblicazione: (2024)
• Implicit Optimization Bias of Next-Token Prediction in Linear Models
  di: Thrampoulidis, Christos
  Pubblicazione: (2024)
• Cautious Next Token Prediction
  di: Wang, Yizhou, et al.
  Pubblicazione: (2025)