Scaling Monosemanticity: Extracting Interpretable Features from Claude 3 Sonnet
Templeton et al., 2024, Scaling Monosemanticity (Transformer Circuits Thread)
Subscriber deck — free during early access
1 cards in “Safety-Relevant Features” · whole deck
Safety-Relevant Features
- What safety-relevant features did Templeton et al. find, and what do they caution?Features for unsafe code, backdoors, bias, sycophancy, deception, power-seeking, secrecy and dangerous content such as bioweapons, which also changed behaviour when clamped. They caution that knowing about lying is not lying: the features’ existence doesn’t show the model is dangerous, nor yet that they are useful for safety.
- Templeton et al., 2024, Scaling Monosemanticity (Transformer Circuits Thread), Safety-Relevant Features
Source
- Adly Templeton, Tom Conerly, Jonathan Marcus, Jack Lindsey, Trenton Bricken, Brian Chen, Adam Pearce, Craig Citro, Emmanuel Ameisen, Andy Jones, Hoagy Cunningham, Nicholas L. Turner, Callum McDougall, Monte MacDiarmid, Alex Tamkin, Esin Durmus, Tristan Hume, Francesco Mosconi, C. Daniel Freeman, Theodore R. Sumers, Edward Rees, Joshua Batson, Adam Jermyn, Shan Carter, Chris Olah and Tom Henighan. 2024. Scaling Monosemanticity: Extracting Interpretable Features from Claude 3 Sonnet. In Transformer Circuits Thread. https://transformer-circuits.pub/2024/scaling-monosemanticity/index.html