Instruction Following on IFEval (avg@4)
89.4IFEval avg@4AutoIF-RLVR
Evaluation Results
| Method | Links | |
|---|---|---|
| AutoIF-RLVR2026.06 | 89.4 | |
| SFT Baseline2026.06 | 84 | |
| QUBRIC2026.06 | 82.3 | |
| Rubric-RLquery=Original Query2026.06 | 80.4 | |
| RM-RLHFlength_penalty=1.02026.06 | 79.6 | |
| RM-RLHFlength_penalty=none2026.06 | 78.7 |