Instruction Following on ImpRIF (test)
80.81CSRImpRIF-32B
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| ImpRIF-32BTraining Stage=SFT+RL2026.02 | 80.81 | 57.62 | |
| ImpRIF-8BTraining Stage=SFT+RL2026.02 | 78.33 | 51.85 | |
| ImpRIF-4BTraining Stage=SFT+RL2026.02 | 75.75 | 49.11 | |
| ImpRIF-32BTraining Stage=RL2026.02 | 75.44 | 43.94 | |
| ImpRIF-32BTraining Stage=SFT2026.02 | 71.86 | 42.03 | |
| ImpRIF-8BTraining Stage=SFT2026.02 | 68.63 | 37.21 | |
| ImpRIF-4BTraining Stage=SFT2026.02 | 67.95 | 36.47 | |
| ImpRIF-8BTraining Stage=RL2026.02 | 66.33 | 32.03 | |
| ImpRIF-4BTraining Stage=RL2026.02 | 62.64 | 28.1 | |
| Qwen3-235B-A22BTraining Stage=Base2026.02 | 59.33 | 21.4 | |
| Qwen3-32BTraining Stage=Base2026.02 | 59.11 | 21.4 | |
| Qwen3-14BTraining Stage=Base2026.02 | 58.14 | 21.65 | |
| Qwen3-8BTraining Stage=Base2026.02 | 55.64 | 19.87 | |
| Qwen3-4BTraining Stage=Base2026.02 | 54 | 17.7 | |
| Ministral3-14BTraining Stage=Base2026.02 | 53.97 | 18.4 | |
| Qwen2.5-72BReasoning Capability=Lacks inference reasoning capabilities2026.02 | 53.85 | 18.13 | |
| Llama3.1-70BReasoning Capability=Lacks inference reasoning capabilities2026.02 | 51 | 16.24 |