Dan Braun

发表论文 7 篇 · 总被引 158 次 · h-index 6

代表论文

• Identifying Functionally Important Features with End-to-End Sparse Dictionary Learning (2024 · Neural Information Processing Systems · 被引 76)
• Towards evaluations-based safety cases for AI scheming (2024 · arXiv.org · 被引 39)
• Interpretability in Parameter Space: Minimizing Mechanistic Description Length with Attribution-based Parameter Decomposition (2025 · arXiv.org · 被引 24)
• A Causal Framework for AI Regulation and Auditing (被引 18)
• Stochastic Parameter Decomposition (2025 · arXiv.org · 被引 14)
• Using Degeneracy in the Loss Landscape for Mechanistic Interpretability (2024 · arXiv.org · 被引 14)