Privacy Expectation Alignment on ConfAIde
68.2Pearson CorrelationQwen3.5-9B
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| Qwen3.5-9BCondition=GRPO2026.04 | 68.2 | 81.8 | |
| Qwen3.5-9BCondition=SFT2026.04 | 67.9 | 80.7 | |
| Qwen3.5-9BCondition=Zero-shot2026.04 | 64.1 | 95.2 | |
| Gemma-3-12BCondition=Zero-shot2026.04 | 61.3 | 74.1 | |
| ContextReasoner-7BCondition=Zero-shot2026.04 | 60.2 | 80.4 | |
| Phi-4Condition=SFT2026.04 | 59.9 | 88.9 | |
| Qwen3.5-4BCondition=Zero-shot2026.04 | 53.7 | 86.3 | |
| Gemma-3-12BCondition=SFT2026.04 | 53.6 | 81.8 | |
| Qwen3.5-4BCondition=SFT2026.04 | 53 | 62.2 | |
| CIRL-7BCondition=Zero-shot2026.04 | 47 | 87 | |
| OpenThinker3-7BCondition=Zero-shot2026.04 | 42.8 | 81.1 | |
| OpenThinker3-7BCondition=SFT2026.04 | 35 | 21.1 | |
| Qwen3.5-2BCondition=Zero-shot2026.04 | 19.2 | 67.4 | |
| Qwen3.5-2BCondition=SFT2026.04 | 13.4 | 35.9 | |
| Phi-4Condition=Zero-shot2026.04 | 9.3 | 91.5 | |
| GPT-OSS-20BCondition=Zero-shot2026.04 | 0 | 98.9 | |
| GPT-OSS-20BCondition=SFT2026.04 | 0 | 97 |