attribution-patching

Gradient-based approximation to activation patching for scalable circuit analysis. Use when activation patching is too slow or when analyzing many components simultaneously.

by ndif-team· Repository·other
Also installable via skills CLI
npx skills add ndif-team/skills/plugins/nnsight/skills/attribution-patching

Source

Path:plugins/nnsight/skills/attribution-patching(main)

Related in other