Fine-grained Knowledge Recall on UnFine-MQ
66.99Bert-ScoreFABLE
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| FABLEModel=Llama3-8B2026.04 | 66.99 | 62.55 | 65.86 | 71.31 | |
| FABLEModel=Qwen2.5-7B2026.04 | 53.16 | 44.01 | 42.13 | 51.1 | |
| UnKEModel=Llama3-8B2026.04 | 30.27 | 39.08 | 37.36 | 49.46 | |
| AnyEditModel=Llama3-8B2026.04 | 29.96 | 38.42 | 36.62 | 49 | |
| AnyEditModel=Qwen2.5-7B2026.04 | 29.48 | 35.66 | 31.59 | 44.43 | |
| UnKEModel=Qwen2.5-7B2026.04 | 29.31 | 34.57 | 29.81 | 42.89 | |
| MEMITModel=Llama3-8B2026.04 | 29.22 | 34.61 | 31.94 | 44.09 | |
| ROMEModel=Qwen2.5-7B2026.04 | 28.66 | 31.73 | 26.47 | 39.39 | |
| MEMITModel=Qwen2.5-7B2026.04 | 28.54 | 33.15 | 26.75 | 40.83 | |
| Pre-editedModel=Qwen2.5-7B2026.04 | 28.28 | 28.81 | 22.94 | 34.91 | |
| FT-LModel=Llama3-8B2026.04 | 28.04 | 24.71 | 18.16 | 29.65 | |
| ROMEModel=Llama3-8B2026.04 | 28.01 | 31.37 | 27.8 | 40.47 | |
| Pre-editedModel=Llama3-8B2026.04 | 27.74 | 30.35 | 25.88 | 37.71 | |
| FT-LModel=Qwen2.5-7B2026.04 | 27.27 | 21.15 | 14.29 | 27.95 |