List your company

SAE Feature Interpretability on Gemma 3 270M SAE Features (Value/Output)

21.82Input Interpretability ScoreQL_KEY
-0.13445.565311.26516.9647May 30, 2026

Evaluation results

MethodLinks
QL_KEY2026.0521.82—1.43
QL_VALUE2026.0510.27—10.63
TCa=102026.0510.05—9.74
TCa=52026.0510.03—8.87
TCa=12026.0510—5.87
ZO2026.055.78—2.75
TL2026.052.23—5.38
LLVALUE2026.052.14—6.13
LLKEY2026.050.71—1.79
Logit Lensvariant=LL_VALUE2026.05—4.1—
Query Lensvariant=QL_VALUE2026.05—5.34—