Jailbreak Success on Experiment 1 (Evaluation)
94.33Success RateBase
Evaluation results
| Method | Links | ||
|---|---|---|---|
| BaseModel=Llama-3.1-8B, Training condition=Base (no FT)2026.06 | 94.33 | — | |
| EmpatheticDialoguesModel=Llama-3.1-8B, Training condition=EmpatheticDialogues2026.06 | 89.67 | 4.66 | |
| Lahnala-styleModel=Llama-3.1-8B, Training condition=Lahnala-style2026.06 | 88.67 | 5.66 | |
| EmpatheticDialoguesModel=Mistral-7B Instruct-v0.3, Training condition=EmpatheticDialogues2026.06 | 88.33 | 28.66 | |
| ESConvModel=Llama-3.1-8B, Training condition=ESConv2026.06 | 84.67 | 9.66 | |
| Lahnala-styleModel=Mistral-7B Instruct-v0.3, Training condition=Lahnala-style2026.06 | 83.67 | 24 | |
| Lahnala-styleModel=SmolLM3-3B, Training condition=Lahnala-style2026.06 | 79.33 | 42.66 | |
| Lahnala-styleModel=Qwen-2.5-7B Instruct, Training condition=Lahnala-style2026.06 | 78 | 38 | |
| EmpatheticDialoguesModel=Qwen-2.5-7B Instruct, Training condition=EmpatheticDialogues2026.06 | 75 | 35 | |
| BaseModel=Mistral-7B Instruct-v0.3, Training condition=Base (no FT)2026.06 | 59.67 | — | |
| EmpatheticDialoguesModel=SmolLM3-3B, Training condition=EmpatheticDialogues2026.06 | 59 | 22.33 | |
| ESConvModel=Qwen-2.5-7B Instruct, Training condition=ESConv2026.06 | 55.33 | 15.33 | |
| PersonaFuse (low A)Model=Llama-3.1-8B, Training condition=PersonaFuse (low A)2026.06 | 51.33 | 43 | |
| ESConvModel=Mistral-7B Instruct-v0.3, Training condition=ESConv2026.06 | 44.33 | 15.34 | |
| PersonaFuse (low A)Model=Mistral-7B Instruct-v0.3, Training condition=PersonaFuse (low A)2026.06 | 43.33 | 16.34 | |
| ESConvModel=SmolLM3-3B, Training condition=ESConv2026.06 | 41 | 4.33 | |
| BaseModel=Qwen-2.5-7B Instruct, Training condition=Base (no FT)2026.06 | 40 | — | |
| BaseModel=SmolLM3-3B, Training condition=Base (no FT)2026.06 | 36.67 | — | |
| PersonaFuse (low A)Model=Qwen-2.5-7B Instruct, Training condition=PersonaFuse (low A)2026.06 | 36 | 4 | |
| PersonaFuse (low A)Model=SmolLM3-3B, Training condition=PersonaFuse (low A)2026.06 | 34.33 | 2.34 |