SOURCE-LINKED INTELLIGENCE
How Far Do SAE Features Persist Across LLM Layers? An Empirical Study with an FFN Pruning Application
Proceedings of the 9th BlackboxNLP Workshop: Analyzing and Interpreting Neural Networks for NLP
Read original source ↗ Open in workspace
- recordType
- paper
- region
- Global
Evidence & attribution
ACL Anthology and the credited paper authors. Bibliographic metadata only; abstracts and full papers are excluded. Third-party materials retain their respective rights.
License: ACL Anthology usage policy (metadata)
First collected: 2026-10-26T00:53:18.250Z. This is not the publication date.