Response reliability estimation on MATH
0.801AUROCLogTokU
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| LogTokUModel=Gemma2026.06 | 0.801 | 0.693 | |
| GLUModel=Gemma2026.06 | 0.799 | 0.699 | |
| GLUModel=Fanar2026.06 | 0.788 | 0.627 | |
| RAUQModel=Fanar2026.06 | 0.779 | 0.624 | |
| RAUQModel=Gemma2026.06 | 0.771 | 0.639 | |
| GLUModel=Qwen2026.06 | 0.767 | 0.537 | |
| RAUQModel=Qwen2026.06 | 0.699 | 0.424 | |
| LogTokUModel=Fanar2026.06 | 0.664 | 0.324 | |
| LogTokUModel=Qwen2026.06 | 0.595 | 0.216 |