General Language Understanding on MMLU (Score)
73.59MMLU ScoreOrigin
Evaluation Results
| Method | Links | |
|---|---|---|
| OriginBase Model=Qwen2.5-7B-Instruct2026.02 | 73.59 | |
| SFTBase Model=Qwen2.5-7B-Instruct2026.02 | 73.48 | |
| RLBase Model=Qwen2.5-7B-Instruct2026.02 | 73.13 | |
| LARFTBase Model=Qwen2.5-7B-Instruct2026.02 | 72.94 | |
| SFT+RLBase Model=Qwen2.5-7B-Instruct2026.02 | 72.67 | |
| PositionIDBase Model=Qwen2.5-7B-Instruct2026.02 | 72.04 | |
| RulerBase Model=Qwen2.5-7B-Instruct2026.02 | 71.02 | |
| OriginBase Model=Llama-3.1-8B-Instruct2026.02 | 68.45 | |
| SFTBase Model=Llama-3.1-8B-Instruct2026.02 | 67.73 | |
| LARFTBase Model=Llama-3.1-8B-Instruct2026.02 | 67.3 | |
| RLBase Model=Llama-3.1-8B-Instruct2026.02 | 66.87 | |
| LARFTBase Model=Qwen2.5-3B-Instruct2026.02 | 66.46 | |
| OriginBase Model=Qwen2.5-3B-Instruct2026.02 | 65.07 | |
| SFT+RLBase Model=Llama-3.1-8B-Instruct2026.02 | 65.05 | |
| SFTBase Model=Qwen2.5-3B-Instruct2026.02 | 65 | |
| RLBase Model=Qwen2.5-3B-Instruct2026.02 | 65 | |
| SFTBase Model=Llama-3.2-3B-Instruct2026.02 | 61.74 | |
| OriginBase Model=Llama-3.2-3B-Instruct2026.02 | 61.5 | |
| RLBase Model=Llama-3.2-3B-Instruct2026.02 | 61.35 | |
| LARFTBase Model=Llama-3.2-3B-Instruct2026.02 | 60.87 | |
| Base Model2026.04 | 60.8 | |
| NGDiff2026.04 | 60.8 | |
| MOLLM2026.04 | 60.8 | |
| Upper Bound2026.04 | 60.7 | |
| SFT+RLBase Model=Llama-3.2-3B-Instruct2026.02 | 60.68 | |
| SimNPOTraining objective=Obj1 only2026.04 | 60.4 | |
| Ensemble Teacher (ET)2026.04 | 60.3 | |
| PositionIDBase Model=Qwen2.5-3B-Instruct2026.02 | 59.97 | |
| MULE2026.04 | 59.2 | |
| FLATTraining objective=Obj1 only2026.04 | 58.92 | |
| SFT+RLBase Model=Qwen2.5-3B-Instruct2026.02 | 54.75 | |
| SFT2026.04 | 50.2 | |
| DUETTeacher model integration=Ensemble Teacher2026.04 | 49.9 | |
| RulerBase Model=Qwen2.5-3B-Instruct2026.02 | 41.94 | |
| PositionIDBase Model=Llama-3.1-8B-Instruct2026.02 | 40.35 | |
| RulerBase Model=Llama-3.1-8B-Instruct2026.02 | 37.15 | |
| GATraining objective=Obj1 only2026.04 | 36.45 | |
| RulerBase Model=Llama-3.2-3B-Instruct2026.02 | 24.05 | |
| PositionIDBase Model=Llama-3.2-3B-Instruct2026.02 | 23.91 |