Takeaways From Our Recent Work on SAE Probing
Josh Engels, Subhash Kantamneni, Senthooran Rajamanoharan, Neel NandaDate not recordedOriginal︎ ↗Archived copy
Examinations
None recorded. An examination is commissioned by a claim’s Steward for what its claim rests on; a review of the document for its own sake is asked for as a request bearing on it.
Text34 passages · 1 marked
We also investigated multi-token probing, where information is aggregated across all tokens in a sequence. When we implemented max-pooling on SAE latents across tokens, the win rate for SAE probes jumped from 2.2% to 19.6% compared to last-token baselines. This initially appeared promising! However, when we implemented an attention-pooled baseline that could also leverage information across tokens, the SAE probe win rate dropped to 8.7%.
Not yet read by this claim’s Steward.
Lineage across 1 claim
Record
- Kind
- blog post
- Authors
- Josh Engels, Subhash Kantamneni, Senthooran Rajamanoharan, Neel Nanda
- Publisher
- Not recorded
- Published
- Not recorded
- DOI
- None
- Link
- https://www.alignmentforum.org/posts/osNKnwiJWHxDYvQTD/takeaways-from-our-recent-work-on-sae-probing
- Archived copy
- https://web.archive.org/web/20260624174828/https://www.alignmentforum.org/posts/osNKnwiJWHxDYvQTD/takeaways-from-our-recent-work-on-sae-probing
- Stored text
- 10,589 characters, retrieved Nov 15, 2026 · 29 sections, 34 passages
- Claims extracted
- Not recorded
- Versions
- One known
- Copies
- None recorded
- Corrections
- None recorded
- Facts checked
- Nov 15, 2026
History 3
- Nov 15, 2026DividedSegmenter · Divided into 62 parts by its own structure
- Nov 15, 2026FactsFacts fetcher · Facts looked up
- Nov 15, 2026Entered the graphEntered the graph; copy stored