History
At L0=160, fully trained SAEs significantly outperform frozen variants in most sparse-probing and causal-editing comparisons.
Claim entered the graph
Nothing has happened to this claim yet beyond its creation: no assessments, contributions, or decisions are on record.