Instruction Following on MultiChallengeBench
57.8ISRImpRIF-32B
Evaluation Results
| Method | Links | |
|---|---|---|
| ImpRIF-32BTraining Stage=SFT+RL2026.02 | 57.8 | |
| ImpRIF-32BTraining Stage=SFT2026.02 | 55.07 | |
| ImpRIF-8BTraining Stage=SFT+RL2026.02 | 48.17 | |
| ImpRIF-8BTraining Stage=SFT2026.02 | 44.28 | |
| Qwen3-32BTraining Stage=Base2026.02 | 44.16 | |
| ImpRIF-32BTraining Stage=RL2026.02 | 43.41 | |
| ImpRIF-4BTraining Stage=SFT+RL2026.02 | 43.13 | |
| Qwen3-14BTraining Stage=Base2026.02 | 41.2 | |
| ImpRIF-4BTraining Stage=SFT2026.02 | 38.6 | |
| Qwen3-235B-A22BTraining Stage=Base2026.02 | 38.45 | |
| Qwen3-8BTraining Stage=Base2026.02 | 36.96 | |
| ImpRIF-8BTraining Stage=RL2026.02 | 36.46 | |
| Qwen3-4BTraining Stage=Base2026.02 | 32.97 | |
| ImpRIF-4BTraining Stage=RL2026.02 | 32.28 | |
| Ministral3-14BTraining Stage=Base2026.02 | 29.68 | |
| Qwen2.5-72BReasoning Capability=Lacks inference reasoning capabilities2026.02 | 20.65 | |
| Llama3.1-70BReasoning Capability=Lacks inference reasoning capabilities2026.02 | 19.93 |