Long-form Writing on WritingBench
88.27ScoreQwen2.5-14B-Writing-RL
Evaluation Results
| Method | Links | |
|---|---|---|
| Qwen2.5-14B-Writing-RLSFT=✓, RL=PPO2025.06 | 88.27 | |
| Qwen2.5-7B-Writing-RLSFT=✓, RL=PPO2025.06 | 87.23 | |
| Llama3.1-8B-Writing-RLSFT=✓, RL=PPO2025.06 | 87.11 | |
| Qwen2.5-7B-Writing-RLSFT=✓, RL=GRPO2025.06 | 86.29 | |
| Qwen2.5-14B-WritingBench-SFTSFT=✓, RL=✗2025.06 | 84.95 | |
| Qwen2.5-7B-Reference-SFTSFT=✓, RL=✗2025.06 | 84.23 | |
| Llama3.1-8B-Reference-SFTSFT=✓, RL=✗2025.06 | 83.98 | |
| Qwen2.5-7B-WritingBench-SFT (24k)SFT=✓, RL=✗2025.06 | 83.76 | |
| Llama3.1-8B-WritingBench-SFTSFT=✓, RL=✗2025.06 | 83.75 | |
| Qwen2.5-7B-WritingBench-SFT (12k)SFT=✓, RL=✗2025.06 | 83.71 | |
| GPT-4oSFT=–, RL=–2025.06 | 83.42 | |
| Longwriter-Zero-32BSFT=✓, RL=GRPO2025.06 | 82.92 | |
| Longwriter-9BSFT=✓, RL=DPO2025.06 | 79.1 | |
| Qwen-PlusSFT=–, RL=–2025.06 | 77.62 | |
| Qwen2.5-7B-InstructSFT=✗, RL=✗2025.06 | 73.16 | |
| Qwen2.5-14B-InstructSFT=✗, RL=✗2025.06 | 72.67 | |
| Llama3.1-8B-InstructSFT=✗, RL=✗2025.06 | 66.56 | |
| Suri-7BSFT=✓, RL=✗2025.06 | 49.7 |