Adding Mistake on FOLIO
0.714AOCFaithfulness Only
Evaluation Results
| Method | Links | |
|---|---|---|
| Faithfulness OnlyBackbone=Qwen3-14B2026.02 | 0.714 | |
| REMULBackbone=Qwen3-14B2026.02 | 0.703 | |
| MAT-SteerBackbone=Qwen3-14B2026.02 | 0.674 | |
| Balanced RewardsBackbone=Qwen3-14B2026.02 | 0.671 | |
| OriginalBackbone=Qwen3-14B2026.02 | 0.667 | |
| Correctness OnlyBackbone=Qwen3-14B2026.02 | 0.648 | |
| Hint OptimizedBackbone=Qwen3-14B2026.02 | 0.641 |