한국어EN
Safety, Alignment & InterpretabilityAnthropic

Scaling Monosemanticity / Golden Gate Claude

Scaling Monosemanticity / Golden Gate Claude

Why it matters

First applied SAEs to a production-grade model and demonstrated feature steering; interpretability moved from observation to intervention

Source — Primary sources first; secondary reports are labelled “Press”

Officialtransformer-circuits.pubhttps://transformer-circuits.pub/2024/scaling-monosemanticity/

Added in Week 35, 2026 · See that week

Same week

xAI closes $6 billion Series BOfficialSimPOPaperHippoRAG publishedPaperCouncil of the EU gives final approval to the AI ActPress

Nearby on the same axis

Frontier Safety Framework v1OfficialOpenAI publishes the first Model SpecOfficialDebating with More Persuasive LLMs (ICML 2024 Best Paper)PaperSleeper AgentsPaper