Holistic Unstructured Text Generation on UnFine-MQ
98.84Bert-ScoreFABLE
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| FABLEModel=Qwen2.5-7B2026.04 | 98.84 | 96.98 | |
| FABLEModel=Llama3-8B2026.04 | 98.71 | 96.59 | |
| UnKEModel=Llama3-8B2026.04 | 98.23 | 95.18 | |
| AnyEditModel=Llama3-8B2026.04 | 97.53 | 93.77 | |
| AnyEditModel=Qwen2.5-7B2026.04 | 96.45 | 90.3 | |
| UnKEModel=Qwen2.5-7B2026.04 | 94.88 | 85.35 | |
| MEMITModel=Qwen2.5-7B2026.04 | 79.44 | 51.71 | |
| MEMITModel=Llama3-8B2026.04 | 78.86 | 50.91 | |
| ROMEModel=Qwen2.5-7B2026.04 | 75.39 | 40.29 | |
| ROMEModel=Llama3-8B2026.04 | 74.04 | 36.1 | |
| Pre-editedModel=Llama3-8B2026.04 | 69.67 | 20.6 | |
| Pre-editedModel=Qwen2.5-7B2026.04 | 69.08 | 20.98 | |
| FT-LModel=Qwen2.5-7B2026.04 | 23.03 | 12.51 | |
| FT-LModel=Llama3-8B2026.04 | 22.65 | 7.4 |