Instruction Following on IFEval (ROCG and ρ)
85.1ROCGFLORA-PMI
Evaluation results
| Method | Links | ||
|---|---|---|---|
| FLORA-PMIModel=G2-9b-it, Eval-time correction=per-task best2026.07 | 85.1 | 0.628 | |
| FLORA-PMIModel=Q3.5-9b, Eval-time correction=per-task best2026.07 | 79.6 | 0.433 | |
| BaseModel=G2-9b-it, Eval-time correction=per-task best2026.07 | 78.2 | 0.355 | |
| RankAlignModel=Q3.5-9b, Eval-time correction=per-task best2026.07 | 75.4 | 0.328 | |
| RankAlignModel=G2-9b-it, Eval-time correction=per-task best2026.07 | 74.9 | 0.299 | |
| SFTModel=G2-9b-it, Eval-time correction=per-task best2026.07 | 74.4 | 0.295 | |
| SFTModel=Q3.5-9b, Eval-time correction=per-task best2026.07 | 73.1 | 0.241 | |
| Consistency FTModel=Q3.5-9b, Eval-time correction=per-task best2026.07 | 69.8 | 0.224 | |
| BaseModel=Q3.5-9b, Eval-time correction=per-task best2026.07 | 69.3 | 0.215 | |
| FLORA-NegModel=Q3.5-9b, Eval-time correction=per-task best2026.07 | 66 | 0.447 | |
| FLORA-NegModel=G2-9b-it, Eval-time correction=per-task best2026.07 | 60.8 | 0.344 | |
| Consistency FTModel=G2-9b-it, Eval-time correction=per-task best2026.07 | 58.7 | 0.008 |