Capability Salience Vector: Fine-grained Alignment of Loss and Capabilities for Downstream Task Scaling Law

Documenti analoghi

• Are Your LLMs Capable of Stable Reasoning?
  di: Liu, Junnan, et al.
  Pubblicazione: (2024)
• Linear Alignment: A Closed-form Solution for Aligning Human Preferences without Tuning and Feedback
  di: Gao, Songyang, et al.
  Pubblicazione: (2024)
• Prism: A Framework for Decoupling and Assessing the Capabilities of VLMs
  di: Qiao, Yuxuan, et al.
  Pubblicazione: (2024)
• How to Set the Learning Rate for Large-Scale Pre-training?
  di: Zhou, Yunhua, et al.
  Pubblicazione: (2026)
• The Fine Line: Navigating Large Language Model Pretraining with Down-streaming Capability Analysis
  di: Yang, Chen, et al.
  Pubblicazione: (2024)