Sequential Model Editing on RIPE (test)
98.7ReliabilityDAFNet
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| DAFNetBackbone=GPT-J (6B), # Editing=102024.05 | 98.7 | 57.6 | 57.6 | 71.3 | |
| DAFNetBackbone=LLAMA2 (7B), # Editing=12024.05 | 97.8 | 66.4 | 72.2 | 78.8 | |
| DAFNetBackbone=LLAMA2 (7B), # Editing=102024.05 | 88.8 | 56.4 | 83.2 | 76.1 | |
| DAFNetBackbone=GPT-J (6B), # Editing=1002024.05 | 78.5 | 38.9 | 64.4 | 60.6 | |
| DAFNetBackbone=LLAMA2 (7B), # Editing=1002024.05 | 57.7 | 41.2 | 87.5 | 62.2 |