Language Model Evaluation on Open LLM Leaderboard v2 (test)
60.84BBHQwen3-8B
Evaluation Results
| Method | Links | |||||||
|---|---|---|---|---|---|---|---|---|
| Qwen3-8BCR=Original, Base Model=Qwen3-8B2026.02 | 60.84 | 36.33 | 39.21 | 52.49 | 47.62 | 43.12 | — | |
| COMPOTCR=0.2, Base Model=Qwen3-8B, Mode=dynamic2026.02 | 50.22 | 31.96 | 33.45 | 12.76 | 38.55 | 41.8 | — | |
| COMPOT†CR=0.2, Base Model=Qwen3-8B, Mode=static2026.02 | 49.33 | 30.54 | 30.46 | 6.72 | 35.92 | 45.11 | — | |
| Mistral+UFTTraining Stage=Instruction-tuning, Model=Mistral+UFT2024.10 | 46.55 | 29.24 | 28.89 | 3.87 | 30.25 | 41.73 | 30.09 | |
| Mistral+SFTTraining Stage=Instruction-tuning, Model=Mistral+SFT2024.10 | 46.04 | 28.72 | 29.5 | 2.66 | 29.35 | 42.94 | 29.87 | |
| Mistral+UFTTraining Stage=Instruction-tuning and Alignment, Model=Mistral+UFT2024.10 | 45.46 | 31.15 | 46.03 | 3.13 | 30.05 | 41.06 | 32.81 | |
| COMPOTCR=0.3, Base Model=Qwen3-8B, Mode=dynamic2026.02 | 44.87 | 28.94 | 31.89 | 2.49 | 32.85 | 40.34 | — | |
| Mistral+SFT+DPOTraining Stage=Instruction-tuning and Alignment, Model=Mistral+SFT+DPO2024.10 | 44.52 | 29.98 | 26.64 | 3.13 | 29.95 | 40.31 | 29.09 | |
| MistralModel=Mistral2024.10 | 44.11 | 29.53 | 23.22 | 2.92 | 30.11 | 41.79 | 28.61 | |
| Mistral+SFT+UNATraining Stage=Instruction-tuning and Alignment, Model=Mistral+SFT+UNA2024.10 | 43.74 | 30.78 | 26.82 | 2.96 | 30.09 | 40.56 | 29.16 | |
| Mistral+SFT+KTOTraining Stage=Instruction-tuning and Alignment, Model=Mistral+SFT+KTO2024.10 | 42.89 | 31 | 25.17 | 2.94 | 30.48 | 40.59 | 28.85 | |
| COMPOT†CR=0.3, Base Model=Qwen3-8B, Mode=static2026.02 | 41.45 | 28.86 | 26.74 | 1.51 | 29.46 | 40.61 | — | |
| SVD-LLMCR=0.2, Base Model=Qwen3-8B2026.02 | 41.03 | 28.27 | 25.66 | 1.06 | 26.3 | 39.81 | — | |
| COMPOTCR=0.4, Base Model=Qwen3-8B, Mode=dynamic2026.02 | 37.51 | 28.1 | 26.14 | 0.98 | 26.44 | 38.89 | — | |
| Qwen3-0.6BCR=Original, Base Model=Qwen3-0.6B2026.02 | 36.85 | 28.27 | 27.58 | 12.69 | 24.26 | 31.35 | — | |
| SVD-LLMCR=0.3, Base Model=Qwen3-8B2026.02 | 34.3 | 25.59 | 22.66 | 1.06 | 18.73 | 41.4 | — | |
| COMPOT†CR=0.4, Base Model=Qwen3-8B, Mode=static2026.02 | 34.13 | 24.75 | 23.14 | 1.06 | 21.23 | 41.27 | — | |
| COMPOT†CR=0.3, Base Model=Qwen3-0.6B, Mode=static2026.02 | 30.83 | 23.49 | 25.06 | 0.23 | 10.85 | 32.8 | — | |
| SVD-LLMCR=0.4, Base Model=Qwen3-8B2026.02 | 30.24 | 22.99 | 22.42 | 0.83 | 11.54 | 37.57 | — | |
| COMPOTCR=0.2, Base Model=Qwen3-0.6B, Mode=dynamic2026.02 | 30.19 | 25.34 | 25.9 | 0.91 | 14.81 | 33.2 | — | |
| COMPOTCR=0.3, Base Model=Qwen3-0.6B, Mode=dynamic2026.02 | 29.99 | 26.59 | 27.7 | 0.68 | 11.43 | 35.71 | — | |
| SVD-LLMCR=0.3, Base Model=Qwen3-0.6B2026.02 | 29.73 | 25.5 | 24.34 | 0 | 11.22 | 33.73 | — | |
| COMPOT†CR=0.4, Base Model=Qwen3-0.6B, Mode=static2026.02 | 29.21 | 25.84 | 24.58 | 0.98 | 10.85 | 36.51 | — | |
| COMPOT†CR=0.2, Base Model=Qwen3-0.6B, Mode=static2026.02 | 29.2 | 27.94 | 27.22 | 1.06 | 12.81 | 35.19 | — | |
| COMPOTCR=0.4, Base Model=Qwen3-0.6B, Mode=dynamic2026.02 | 29.06 | 23.74 | 26.62 | 0.38 | 10.66 | 33.73 | — | |
| SVD-LLMCR=0.4, Base Model=Qwen3-0.6B2026.02 | 29.01 | 24.33 | 22.3 | 0 | 10.98 | 36.24 | — | |
| SVD-LLMCR=0.2, Base Model=Qwen3-0.6B2026.02 | 28.95 | 24.75 | 25.3 | 0 | 11.29 | 35.58 | — | |
| SPIN Iter-3Model=Llama3.1-8B2026.05 | 28.46 | 3.36 | 44.13 | 5.21 | 22.92 | 12.96 | 19.51 | |
| SPIN Iter-4Model=Llama3.1-8B2026.05 | 28.36 | 3.8 | 44.01 | 4.98 | 23.02 | 13.44 | 19.6 | |
| SPIN Iter-2Model=Llama3.1-8B2026.05 | 28.17 | 4.81 | 43.21 | 5.14 | 22.93 | 13.28 | 19.59 | |
| TPAW Iter-1Model=Llama3.1-8B2026.05 | 28.16 | 6.15 | 38.76 | 5.21 | 23.17 | 13.65 | 19.18 | |
| TPAW Iter-4Model=Llama3.1-8B2026.05 | 28.14 | 6.15 | 45.12 | 6.34 | 23.56 | 15.73 | 20.84 | |
| SPIN Iter-1Model=Llama3.1-8B2026.05 | 28.05 | 6.26 | 42.91 | 4.83 | 22.84 | 13.17 | 19.68 | |
| TPAW Iter-2Model=Llama3.1-8B2026.05 | 27.85 | 6.38 | 42.43 | 6.19 | 23.02 | 16.22 | 20.35 | |
| TPAW Iter-3Model=Llama3.1-8B2026.05 | 27.82 | 6.26 | 45.04 | 5.89 | 23.34 | 16.16 | 20.75 | |
| SFTModel=Llama3.1-8B2026.05 | 26.28 | 6.15 | 36.34 | 4.61 | 21.01 | 11.77 | 17.69 | |
| DPOModel=Llama3.1-8B2026.05 | 25.83 | 6.6 | 36.83 | 4.53 | 21.91 | 11.78 | 17.91 | |
| DPOModel=Qwen2.5-1.5B2026.05 | 16.25 | 2.24 | 31.79 | 9.89 | 18.24 | 2.57 | 13.5 | |
| TPAW Iter-1Model=Qwen2.5-1.5B2026.05 | 15.6 | 3.24 | 32.35 | 8.69 | 17.65 | 5.66 | 13.87 | |
| TPAW Iter-4Model=Qwen2.5-1.5B2026.05 | 15.51 | 4.36 | 36.04 | 9.37 | 17.82 | 4.73 | 14.64 | |
| SPIN Iter-1Model=Qwen2.5-1.5B2026.05 | 15.47 | 4.03 | 32.44 | 8.23 | 17.57 | 5.55 | 13.88 | |
| TPAW Iter-2Model=Qwen2.5-1.5B2026.05 | 15.37 | 4.36 | 35.52 | 8.69 | 17.72 | 5.98 | 14.61 | |
| SPIN Iter-2Model=Qwen2.5-1.5B2026.05 | 15.22 | 3.69 | 33.97 | 8.99 | 17.52 | 5.65 | 14.17 | |
| SFTModel=Qwen2.5-1.5B2026.05 | 15.15 | 3.47 | 31.73 | 6.65 | 17.38 | 6.02 | 13.4 | |
| TPAW Iter-3Model=Qwen2.5-1.5B2026.05 | 15.03 | 4.14 | 36.1 | 10.2 | 18.03 | 5.43 | 14.82 | |
| SPIN Iter-3Model=Qwen2.5-1.5B2026.05 | 14.44 | 4.47 | 35.34 | 9.06 | 17.58 | 5.16 | 14.34 | |
| SPIN Iter-4Model=Qwen2.5-1.5B2026.05 | 14.33 | 4.25 | 33.82 | 9.21 | 17.54 | 5.7 | 14.14 | |
| CodePrefBase Model=OLMo-2-7B-SFT, DPO Training Dataset=CodePref2025.11 | — | — | — | — | — | — | 34.41 | |
| HelpSteerBase Model=OLMo-2-7B-SFT, DPO Training Dataset=HelpSteer2025.11 | — | — | — | — | — | — | 34.3 | |
| ORPOBase Model=OLMo-2-7B-SFT, DPO Training Dataset=ORPO2025.11 | — | — | — | — | — | — | 35.66 | |
| SFTBase Model=OLMo-2-7B-SFT, DPO Training Dataset=SFT2025.11 | — | — | — | — | — | — | 33.99 | |
| TuluDPOBase Model=OLMo-2-7B-SFT, DPO Training Dataset=TuluDPO2025.11 | — | — | — | — | — | — | 36.35 | |
| UltraFBBase Model=OLMo-2-7B-SFT, DPO Training Dataset=UltraFB2025.11 | — | — | — | — | — | — | 35.54 | |
| UM-170kBase Model=OLMo-2-7B-SFT, DPO Training Dataset=UM-170k2025.11 | — | — | — | — | — | — | 36.9 | |
| UM-187kBase Model=OLMo-2-7B-SFT, DPO Training Dataset=UM-187k2025.11 | — | — | — | — | — | — | 37.45 | |
| UM-190kBase Model=OLMo-2-7B-SFT, DPO Training Dataset=UM-190k2025.11 | — | — | — | — | — | — | 37.63 |