Distribution Alignment on UK Dataset Even Setting (test)
0.05Alignment ErrorSteering Token Calibration
Evaluation Results
| Method | Links | |
|---|---|---|
| Steering Token CalibrationModel=Qwen2.5-7B2026.04 | 0.05 | |
| Steering Token CalibrationModel=Qwen2.5-1.5B2026.04 | 0.061 | |
| IFTModel=Llama-3.2-1B2026.04 | 0.085 | |
| IFTModel=Llama-3.1-8B2026.04 | 0.101 | |
| Steering Token CalibrationModel=Llama-3.2-1B2026.04 | 0.12 | |
| Steering Token CalibrationModel=Llama-3.1-8B2026.04 | 0.134 | |
| IFTModel=Qwen2.5-1.5B2026.04 | 0.154 | |
| IFTModel=Qwen2.5-7B2026.04 | 0.189 | |
| DPOModel=Qwen2.5-7B2026.04 | 0.5 | |
| DPOModel=Qwen2.5-1.5B2026.04 | 0.5 | |
| DPOModel=Llama-3.1-8B2026.04 | 0.5 | |
| DPOModel=Llama-3.2-1B2026.04 | 0.5 |