Jacobian Scopes: token-level causal attributions in LLMs

Documenti analoghi

• LLMs learn governing principles of dynamical systems, revealing an in-context neural scaling law
  di: Liu, Toni J. B., et al.
  Pubblicazione: (2024)
• Density estimation with LLMs: a geometric investigation of in-context learning trajectories
  di: Liu, Toni J. B., et al.
  Pubblicazione: (2024)
• What's in a prompt? Language models encode literary style in prompt embeddings
  di: Sarfati, Raphaël, et al.
  Pubblicazione: (2025)
• Lines of Thought in Large Language Models
  di: Sarfati, Raphaël, et al.
  Pubblicazione: (2024)
• Text-Trained LLMs Can Zero-Shot Extrapolate PDE Dynamics, Revealing a Three-Stage In-Context Learning Mechanism
  di: Bao, Jiajun, et al.
  Pubblicazione: (2025)