Pronoun Steering on Pronoun Change (test)
98.66SEKASEKA
Evaluation results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| SEKAModel=QWEN3-14B2026.03 | 98.66 | — | — | — | — | |
| SEKAModel=QWEN3-8B2026.03 | 98.56 | — | — | — | — | |
| SEKAModel=GEMMA3-12B2026.03 | 97.7 | — | — | — | — | |
| SEKAModel=QWEN3-4B2026.03 | 95.18 | — | — | — | — | |
| SEKAModel=GEMMA3-4B2026.03 | 81.53 | — | — | — | — | |
| AdaSEKAModel=QWEN3-4B2026.03 | — | — | — | 94.54 | 1.64 | |
| AdaSEKAModel=QWEN3-8B2026.03 | — | — | — | 99.68 | 0.2 | |
| AdaSEKAModel=QWEN3-14B2026.03 | — | — | — | 99.88 | 0.22 | |
| AdaSEKAModel=GEMMA3-4B2026.03 | — | — | — | 93.76 | 3.6 | |
| AdaSEKAModel=GEMMA3-12B2026.03 | — | — | — | 96.88 | 1.06 | |
| PRISM-∆Model=QWEN3-4B2026.03 | — | 96.18 | — | — | — | |
| PRISM-∆Model=QWEN3-8B2026.03 | — | 99.4 | — | — | — | |
| PRISM-∆Model=QWEN3-14B2026.03 | — | 99.66 | — | — | — | |
| PRISM-∆Model=GEMMA3-4B2026.03 | — | 89.08 | — | — | — | |
| PRISM-∆Model=GEMMA3-12B2026.03 | — | 97.94 | — | — | — | |
| PRISM-∆VModel=QWEN3-4B2026.03 | — | — | 96.06 | — | — | |
| PRISM-∆VModel=QWEN3-8B2026.03 | — | — | 99.48 | — | — | |
| PRISM-∆VModel=QWEN3-14B2026.03 | — | — | 99.66 | — | — | |
| PRISM-∆VModel=GEMMA3-4B2026.03 | — | — | 90.16 | — | — | |
| PRISM-∆VModel=GEMMA3-12B2026.03 | — | — | 94.54 | — | — |