Long-form text generation on LongBench Write-en 1.0 (test)
91.8SqGPT-4o
Evaluation Results
| Method | Links | |
|---|---|---|
| GPT-4o2025.02 | 91.8 | |
| GPT-4o mini2025.02 | 90.3 | |
| LongWriter-Qwen + LongDPOBackbone=Qwen, Optimization=LongDPO2025.02 | 88.6 | |
| Mistral-Large-Instruct2025.02 | 88.3 | |
| LongWriter-Llama + LongDPOBackbone=Llama, Optimization=LongDPO2025.02 | 88.2 | |
| Claude 3.5 Sonnet2025.02 | 87.7 | |
| GPT-4 Turbo2025.02 | 86.6 | |
| GLM-4-9B-chat2025.02 | 85.5 | |
| LongWriter-LlamaBackbone=Llama2025.02 | 82.2 | |
| Llama-3.1-70B-InstructParameters=70B2025.02 | 80.3 | |
| Llama-3.1-8B-InstructParameters=8B2025.02 | 70.6 | |
| Suri-I-ORPO2025.02 | 53.5 |