Early-Signal Detector

method · first seen 4 Oct 2026

safety feature discovery in transcoder features

Also: CSFD

A filtering approach that identifies candidate safety-relevant features from transcoder features without exhaustive ablation.

Mentions per week

0119 May11 Aug10 NovPapersCodeDiscussion

PapersCodeDiscussion

Earliest sources

Latest sources

Mentioned together with

safety refusal in language modelscircuit analysis