Instruction Following on AlpacaEval German
65.2Win RateCLO
Evaluation results
| Method | Links | ||
|---|---|---|---|
| CLOModel=Qwen-2.5-3B, Eval Lang=English2025.05 | 65.2 | 12.3 | |
| CLOModel=Llama-2-7B, Eval Lang=English2025.05 | 61.2 | 12.8 | |
| CLOModel=Llama-2-7B, Eval Lang=Target2025.05 | 59.5 | 9.2 | |
| CLOModel=Llama-2-13B, Eval Lang=English2025.05 | 58.5 | 5.5 | |
| CLOModel=Llama-3-8B, Eval Lang=English2025.05 | 55.5 | 5.4 | |
| CLOModel=Llama-3-8B, Eval Lang=Target2025.05 | 54.6 | 2.9 | |
| CLOModel=Llama-2-13B, Eval Lang=Target2025.05 | 53.7 | 3.2 | |
| SFT+DPOModel=Llama-2-13B, Eval Lang=English2025.05 | 53 | — | |
| SFT+DPOModel=Qwen-2.5-3B, Eval Lang=English2025.05 | 52.9 | — | |
| SFT+DPOModel=Llama-3-8B, Eval Lang=Target2025.05 | 51.7 | — | |
| CLOModel=Mistral-7B-v0.1, Eval Lang=Target2025.05 | 50.8 | 0.9 | |
| CLOModel=Mistral-7B-v0.1, Eval Lang=English2025.05 | 50.6 | 2 | |
| SFT+DPOModel=Llama-2-13B, Eval Lang=Target2025.05 | 50.5 | — | |
| CLOModel=Qwen-2.5-3B, Eval Lang=Target2025.05 | 50.5 | 2 | |
| SFT+DPOModel=Llama-2-7B, Eval Lang=Target2025.05 | 50.3 | — | |
| SFT+DPOModel=Llama-3-8B, Eval Lang=English2025.05 | 50.1 | — | |
| SFT+DPOModel=Mistral-7B-v0.1, Eval Lang=Target2025.05 | 49.9 | — | |
| SFT+DPOModel=Mistral-7B-v0.1, Eval Lang=English2025.05 | 48.6 | — | |
| SFT+DPOModel=Qwen-2.5-3B, Eval Lang=Target2025.05 | 48.5 | — | |
| SFT+DPOModel=Llama-2-7B, Eval Lang=English2025.05 | 48.4 | — |