Response Reliability Estimation on TruthfulQA
0.68AUROCLogTokU
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| LogTokUModel=Qwen2026.06 | 0.68 | 0.377 | |
| RAUQModel=Fanar2026.06 | 0.641 | 0.282 | |
| GLUModel=Fanar2026.06 | 0.641 | 0.307 | |
| GLUModel=Qwen2026.06 | 0.616 | 0.195 | |
| RAUQModel=Qwen2026.06 | 0.582 | 0.063 | |
| LogTokUModel=Fanar2026.06 | 0.579 | 0.192 | |
| GLUModel=Gemma2026.06 | 0.548 | 0.087 | |
| LogTokUModel=Gemma2026.06 | 0.531 | 0.115 | |
| RAUQModel=Gemma2026.06 | 0.466 | 0.094 |