From Reasoning to Answer: Empirical, Attention-Based and Mechanistic Insights into Distilled DeepSeek R1 Models

Documenti analoghi

• Contrastive Attribution in the Wild: An Interpretability Analysis of LLM Failures on Realistic Benchmarks
  di: Tan, Rongyuan, et al.
  Pubblicazione: (2026)
• LLM Reasoning as Trajectories: Step-Specific Representation Geometry and Correctness Signals
  di: Sun, Lihao, et al.
  Pubblicazione: (2026)
• MEETING DELEGATE: Benchmarking LLMs on Attending Meetings on Our Behalf
  di: Hu, Lingxiang, et al.
  Pubblicazione: (2025)
• Navigating the Unknown: A Chat-Based Collaborative Interface for Personalized Exploratory Tasks
  di: Peng, Yingzhe, et al.
  Pubblicazione: (2024)
• Distill Not Only Data but Also Rewards: Can Smaller Language Models Surpass Larger Ones?
  di: Zhang, Yudi, et al.
  Pubblicazione: (2025)