Creative Writing on Creative Writing (Win Rate)
58.6Win RateMin-k
Evaluation Results
| Method | Links | |
|---|---|---|
| Min-kModel=LLaMA3-8B-Instruct, Temperature (T)=1.02026.04 | 58.6 | |
| Min-kModel=Qwen3-4B-Instruct, Temperature (T)=1.02026.04 | 56 | |
| Min-kModel=Qwen3-4B-Instruct, Temperature (T)=3.02026.04 | 55.4 | |
| Top-nσModel=Qwen3-4B-Instruct, Temperature (T)=1.02026.04 | 54.8 | |
| Top-pModel=Qwen3-4B-Instruct, Temperature (T)=1.02026.04 | 54.6 | |
| Top-nσModel=LLaMA3-8B-Instruct, Temperature (T)=5.02026.04 | 54.2 | |
| Min-kModel=LLaMA3-8B-Instruct, Temperature (T)=3.02026.04 | 53.6 | |
| Min-pModel=LLaMA3-8B-Instruct, Temperature (T)=1.02026.04 | 53.4 | |
| Min-kModel=LLaMA3-8B-Instruct, Temperature (T)=5.02026.04 | 53.2 | |
| Min-kModel=LLaMA3-8B-Instruct, Temperature (T)=10.02026.04 | 52.8 | |
| Min-kModel=Qwen3-4B-Instruct, Temperature (T)=10.02026.04 | 52.8 | |
| η-samplingModel=LLaMA3-8B-Instruct, Temperature (T)=1.02026.04 | 52.6 | |
| Top-nσModel=LLaMA3-8B-Instruct, Temperature (T)=1.02026.04 | 52.4 | |
| Min-kModel=Qwen3-4B-Instruct, Temperature (T)=5.02026.04 | 52.2 | |
| Min-pModel=LLaMA3-8B-Instruct, Temperature (T)=3.02026.04 | 51.6 | |
| Top-nσModel=LLaMA3-8B-Instruct, Temperature (T)=3.02026.04 | 51.4 | |
| Top-nσModel=Qwen3-4B-Instruct, Temperature (T)=10.02026.04 | 51.2 | |
| Top-kModel=LLaMA3-8B-Instruct, Temperature (T)=1.02026.04 | 50.8 | |
| Top-nσModel=Qwen3-4B-Instruct, Temperature (T)=3.02026.04 | 50.8 | |
| Top-nσModel=Qwen3-4B-Instruct, Temperature (T)=5.02026.04 | 50.6 | |
| Top-kModel=Qwen3-4B-Instruct, Temperature (T)=1.02026.04 | 50 | |
| Top-nσModel=LLaMA3-8B-Instruct, Temperature (T)=10.02026.04 | 50 | |
| η-samplingModel=Qwen3-4B-Instruct, Temperature (T)=1.02026.04 | 49.8 | |
| Top-pModel=LLaMA3-8B-Instruct, Temperature (T)=1.02026.04 | 49.6 | |
| Min-pModel=Qwen3-4B-Instruct, Temperature (T)=1.02026.04 | 49.4 | |
| MirostatModel=Qwen3-4B-Instruct, Temperature (T)=1.02026.04 | 48.4 | |
| MirostatModel=LLaMA3-8B-Instruct, Temperature (T)=1.02026.04 | 48.2 | |
| Min-pModel=Qwen3-4B-Instruct, Temperature (T)=3.02026.04 | 46.6 | |
| MirostatModel=Qwen3-4B-Instruct, Temperature (T)=3.02026.04 | 18 | |
| Top-kModel=Qwen3-4B-Instruct, Temperature (T)=3.02026.04 | 13.8 | |
| Top-kModel=LLaMA3-8B-Instruct, Temperature (T)=3.02026.04 | 1.4 | |
| MirostatModel=LLaMA3-8B-Instruct, Temperature (T)=3.02026.04 | 1.4 | |
| Top-pModel=Qwen3-4B-Instruct, Temperature (T)=3.02026.04 | 0.4 | |
| η-samplingModel=Qwen3-4B-Instruct, Temperature (T)=3.02026.04 | 0.2 | |
| Top-pModel=LLaMA3-8B-Instruct, Temperature (T)=3.02026.04 | 0 | |
| η-samplingModel=LLaMA3-8B-Instruct, Temperature (T)=3.02026.04 | 0 |