Textual Understanding on AlpacaEval 2
54.7Length Constraint (LC)Uni-DPO
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| Uni-DPOModel=Gemma-2-9B-IT2025.06 | 54.7 | 52.7 | |
| SimPOModel=Gemma-2-9B-IT2025.06 | 53.2 | 48 | |
| SFTModel=Gemma-2-9B-IT2025.06 | 52.5 | 40.5 | |
| Uni-DPOModel=Llama3-8B Instruct2025.06 | 47.2 | 44.2 | |
| Uni-DPO v0.2Model=Llama3-8B Instruct2025.06 | 46.8 | 42.6 | |
| SimPOModel=Llama3-8B Instruct2025.06 | 44.7 | 40.4 | |
| DPOModel=Llama3-8B Instruct2025.06 | 43.7 | 41.7 | |
| SimPO v0.2Model=Llama3-8B Instruct2025.06 | 41.2 | 37.1 | |
| SFTModel=Llama3-8B Instruct2025.06 | 28.5 | 28.4 | |
| Uni-DPOModel=Qwen2.5-7B2025.06 | 25.8 | 24.7 | |
| Uni-DPOModel=Llama3-8B Base2025.06 | 23.8 | 20.5 | |
| Uni-DPO v0.2Model=Llama3-8B Base2025.06 | 22.8 | 24.1 | |
| SimPOModel=Qwen2.5-7B2025.06 | 20.9 | 18.4 | |
| SimPOModel=Llama3-8B Base2025.06 | 19.4 | 18.1 | |
| DPOModel=Llama3-8B Base2025.06 | 15.5 | 13.3 | |
| SimPO v0.2Model=Llama3-8B Base2025.06 | 15.5 | 18.5 | |
| SFTModel=Llama3-8B Base2025.06 | 5.7 | 3.7 |