Long-form writing on LongBench-Write-en ([0, 500) context length)
92.1Sl ScoreGPT-4o
Evaluation results
| Method | Links | ||
|---|---|---|---|
| GPT-4o2025.02 | 92.1 | 93.1 | |
| Qwen-2.5-14Btuning=LongDPO2025.02 | 91.75 | 91.25 | |
| Qwen-2.5-14Btuning=DPO2025.02 | 91.72 | 90.53 | |
| Llama3.1-70B-instruct2025.02 | 90.8 | 84.8 | |
| Llama3.1-8B-instruct2025.02 | 89.7 | 84.6 | |
| QWQ2025.02 | 89.1 | 94.58 | |
| Qwen-2.5-14Btuning=Base2025.02 | 88.39 | 89.77 |