Activation Steering
Benchmarks
Dataset NameSOTA methodMetricTrendResultsLast Updated
46Avg Activation Change
27
Apr 7, 2026
1.179Granularity
20
May 19, 2026
94.85ASR Accuracy
20
Mar 30, 2026
0.871Steering Score
18
May 6, 2026
56Avg. Activation Change
18
Apr 7, 2026
1.12Steering Score
17
May 6, 2026
86.8Edit Reference Rate
12
May 21, 2026
44Input Fidelity
7
May 12, 2026
46.7Input Fidelity
7
May 12, 2026
0.79Steering Effect (Delta s)
5
May 12, 2026
0Steering Effect (Delta s)
5
May 12, 2026
1.13Steering Effect (Delta s)
5
May 12, 2026
0.49Steering Effect (Delta s)
5
May 12, 2026
44Steering Effect (Delta s)
5
May 12, 2026
0.5Steering Effect (Delta s)
5
May 12, 2026