List your company

LLM Steering on BASE→USER.POS (out-of-distribution)

2.535SteerabilityL2S
1.08421.460851.83752.21415Apr 4, 2026

Evaluation results

MethodLinks
L2SModel=Llama-2-7B-Chat, Layer Selection Strategy=W2S2026.042.53595
L2SModel=Llama-2-7B-Chat, Layer Selection Strategy=Fixed Layer2026.042.10292.1
CAAModel=Llama-2-7B-Chat, Layer Selection Strategy=W2S2026.041.67478.4
L2SModel=Qwen1.5-14B-Chat, Layer Selection Strategy=W2S2026.041.63478.5
CAAModel=Qwen1.5-14B-Chat, Layer Selection Strategy=W2S2026.041.45480.3
L2SModel=Qwen1.5-14B-Chat, Layer Selection Strategy=Fixed Layer2026.041.43477.9
CAAModel=Qwen1.5-14B-Chat, Layer Selection Strategy=Fixed Layer2026.041.29576.8
CAAModel=Llama-2-7B-Chat, Layer Selection Strategy=Fixed Layer2026.041.1475.3