Activation Steering on TONEBANK
46Avg Activation ChangeK-Steer
Evaluation Results
| Method | Links | |
|---|---|---|
| K-SteerK=1, Model=Mistral2025.05 | 46 | |
| K-SteerK=1, Model=Olmo2025.05 | 37 | |
| CAAK=1, Model=Mistral2025.05 | 24 | |
| DCTK=1, Model=Mistral2025.05 | 24 | |
| K-SteerK=2, Model=Olmo2025.05 | 21 | |
| DCTK=1, Model=Olmo2025.05 | 18 | |
| K-SteerK=1, Model=Llama2025.05 | 17 | |
| K-SteerK=2, Model=Llama2025.05 | 17 | |
| DCTK=2, Model=Olmo2025.05 | 15 | |
| DCTK=1, Model=Llama2025.05 | 13 | |
| CAAK=2, Model=Olmo2025.05 | 13 | |
| DCTK=2, Model=Llama2025.05 | 13 | |
| CAAK=1, Model=Llama2025.05 | 12 | |
| CAAK=2, Model=Llama2025.05 | 12 | |
| CAAK=2, Model=Mistral2025.05 | 12 | |
| CAAK=1, Model=Olmo2025.05 | 11 | |
| K-SteerK=2, Model=Mistral2025.05 | 9 | |
| K-SteerK=3, Model=Olmo2025.05 | 9 | |
| DCTK=3, Model=Mistral2025.05 | 7 | |
| K-SteerK=3, Model=Llama2025.05 | 7 | |
| K-SteerK=3, Model=Mistral2025.05 | 7 | |
| DCTK=2, Model=Mistral2025.05 | 5 | |
| CAAK=3, Model=Olmo2025.05 | 4 | |
| DCTK=3, Model=Llama2025.05 | 4 | |
| CAAK=3, Model=Mistral2025.05 | 3 | |
| DCTK=3, Model=Olmo2025.05 | 1 | |
| CAAK=3, Model=Llama2025.05 | 0 |