List your company

SAE steering on Neutral prefixes dataset

0.35Concept Improvement (C)Dream-Mask
0.01720.10360.190.2764Feb 5, 2026

Evaluation results

MethodLinks
Dream-MaskLayer=L23, Layer Category=Deep, L0=802026.020.35-0.350.24
Dream-UnmaskLayer=L27, Layer Category=Deep, L0=802026.020.330.180.38
Dream-MaskLayer=L27, Layer Category=Deep, L0=802026.020.290.010.29
Qwen-2.5-7BLayer=L23, Layer Category=Deep, L0=802026.020.28-0.430.15
Dream-UnmaskLayer=L23, Layer Category=Deep, L0=802026.020.26-0.330.16
LLaMA-3-8BLayer=L14, Layer Category=Middle, L0=802026.020.25-0.670.05
LLaDA-UnmaskLayer=L23, Layer Category=Deep, L0=802026.020.25-0.160.2
Qwen-2.5-7BLayer=L14, Layer Category=Middle, L0=802026.020.22-0.260.14
Dream-MaskLayer=L14, Layer Category=Middle, L0=802026.020.22-0.370.11
LLaDA-MaskLayer=L23, Layer Category=Deep, L0=802026.020.21-0.120.17
Dream-UnmaskLayer=L14, Layer Category=Middle, L0=802026.020.2-0.640.01
Dream-UnmaskLayer=L10, Layer Category=Middle, L0=802026.020.19-0.310.1
Qwen-2.5-7BLayer=L5, Layer Category=Shallow, L0=802026.020.18-0.340.08
Qwen-2.5-7BLayer=L10, Layer Category=Middle, L0=802026.020.18-0.350.08
LLaMA-3-8BLayer=L5, Layer Category=Shallow, L0=802026.020.18-2.14-0.46
Dream-UnmaskLayer=L5, Layer Category=Shallow, L0=802026.020.17-0.170.12
Dream-MaskLayer=L10, Layer Category=Middle, L0=802026.020.17-0.120.13
LLaMA-3-8BLayer=L23, Layer Category=Deep, L0=802026.020.17-0.440.04
Qwen-2.5-7BLayer=L1, Layer Category=Shallow, L0=802026.020.16-0.390.04
LLaDA-UnmaskLayer=L5, Layer Category=Shallow, L0=802026.020.16-0.130.13
LLaDA-UnmaskLayer=L27, Layer Category=Deep, L0=802026.020.16-0.260.08
LLaDA-MaskLayer=L14, Layer Category=Middle, L0=802026.020.16-0.060.14
LLaMA-3-8BLayer=L10, Layer Category=Middle, L0=802026.020.15-1.23-0.22
LLaDA-MaskLayer=L27, Layer Category=Deep, L0=802026.020.15-0.020.15
Dream-MaskLayer=L1, Layer Category=Shallow, L0=802026.020.14-0.110.11
LLaDA-UnmaskLayer=L1, Layer Category=Shallow, L0=802026.020.14-0.050.13
LLaDA-MaskLayer=L10, Layer Category=Middle, L0=802026.020.140.040.15
Dream-UnmaskLayer=L1, Layer Category=Shallow, L0=802026.020.13-0.080.1
Dream-MaskLayer=L5, Layer Category=Shallow, L0=802026.020.13-0.090.11
LLaMA-3-8BLayer=L27, Layer Category=Deep, L0=802026.020.13-0.090.1
LLaDA-UnmaskLayer=L10, Layer Category=Middle, L0=802026.020.130.020.13
LLaDA-UnmaskLayer=L14, Layer Category=Middle, L0=802026.020.13-0.040.11
LLaDA-MaskLayer=L1, Layer Category=Shallow, L0=802026.020.1300.13
LLaDA-MaskLayer=L5, Layer Category=Shallow, L0=802026.020.1100.11
LLaMA-3-8BLayer=L1, Layer Category=Shallow, L0=802026.020.08-7.89-2.29
Qwen-2.5-7BLayer=L27, Layer Category=Deep, L0=802026.020.03-0.19-0.02