LLM Steering on BASE→SYS.POS (out-of-distribution)
2.802SteerabilityL2S
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| L2SModel=Llama-2-7B-Chat, Layer Selection Strategy=W2S2026.04 | 2.802 | 97 | |
| L2SModel=Llama-2-7B-Chat, Layer Selection Strategy=Fixed Layer2026.04 | 2.456 | 95.5 | |
| L2SModel=Qwen1.5-14B-Chat, Layer Selection Strategy=W2S2026.04 | 2.192 | 86.6 | |
| L2SModel=Qwen1.5-14B-Chat, Layer Selection Strategy=Fixed Layer2026.04 | 1.829 | 85.7 | |
| CAAModel=Llama-2-7B-Chat, Layer Selection Strategy=W2S2026.04 | 1.821 | 79 | |
| CAAModel=Qwen1.5-14B-Chat, Layer Selection Strategy=W2S2026.04 | 1.685 | 83.3 | |
| CAAModel=Qwen1.5-14B-Chat, Layer Selection Strategy=Fixed Layer2026.04 | 1.427 | 77.9 | |
| CAAModel=Llama-2-7B-Chat, Layer Selection Strategy=Fixed Layer2026.04 | 1.422 | 75 |