Open-ended generation on WritingPrompts (BERTScore, ROUGE-L)
87.8BERTScoreReLAR
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| ReLAR2026.06 | 87.8 | 11.47 | |
| Mistral-7B-Instruct2026.06 | 87.1 | 9.28 | |
| Llama-3-8B-Instruct2026.06 | 86.9 | 8.74 | |
| Qwen2.5-7B2026.06 | 86.4 | 7.94 | |
| Mistral-7B-v0.32026.06 | 85.8 | 6.83 | |
| Gemma-7B2026.06 | 85.3 | 6.17 | |
| LLaMA-2-7B2026.06 | 84.7 | 5.12 | |
| Falcon-7B2026.06 | 84.3 | 4.63 |