Intent Leakage on LawQA
25.43ASR@1PP-TS
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| PP-TSTrusted Local Model=Llama-3.1-8B, Untrusted External Model=GPT-4o mini2026.04 | 25.43 | 48.46 | 37.92 | |
| Soft-POTrusted Local Model=Llama-3.1-8B, Untrusted External Model=GPT-4o mini2026.04 | 24.68 | 47.64 | 37.09 | |
| IOITrusted Local Model=Llama-3.1-8B, Untrusted External Model=GPT-4o mini2026.04 | 23.98 | 46.71 | 36.27 | |
| Hard-POTrusted Local Model=Llama-3.1-8B, Untrusted External Model=GPT-4o mini2026.04 | 23.19 | 45.85 | 35.61 | |
| GTKATrusted Local Model=Llama-3.1-8B, Untrusted External Model=GPT-4o mini2026.04 | 22.55 | 44.93 | 34.68 |