HoPE: Hyperbolic Rotary Positional Encoding for Stable Long-Range Dependency Modeling in Large Language Models

Documenti analoghi

• HoPE: A Novel Positional Encoding Without Long-Term Decay for Enhanced Context Awareness and Extrapolation
  di: Chen, Yuhan, et al.
  Pubblicazione: (2024)
• R-Capsule: Compressing High-Level Plans for Efficient Large Language Model Reasoning
  di: Shan, Hongyu, et al.
  Pubblicazione: (2025)
• HoPE: Hybrid of Position Embedding for Long Context Vision-Language Models
  di: Li, Haoran, et al.
  Pubblicazione: (2025)
• PoPE: Legendre Orthogonal Polynomials Based Position Encoding for Large Language Models
  di: Aggarwal, Arpit
  Pubblicazione: (2024)
• VRoPE: Rotary Position Embedding for Video Large Language Models
  di: Liu, Zikang, et al.
  Pubblicazione: (2025)