Faithfulness Evaluation on Fakepedia Conflicting long-context
849.1AOPCcompATTN
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| ATTNModel=Qwen2.5-7B2025.10 | 849.1 | 1,028.22 | |
| FAModel=Qwen2.5-7B2025.10 | 843.7 | 1,047.21 | |
| IGModel=Qwen2.5-7B2025.10 | 834.6 | 1,052.99 | |
| MechLightModel=Qwen2.5-7B2025.10 | 819 | 1,019.6 | |
| FAModel=Pythia-6.9B2025.10 | 815.6 | 1,100.21 | |
| IGModel=Pythia-6.9B2025.10 | 811.1 | 1,111.49 | |
| MechLightModel=Pythia-6.9B2025.10 | 691.1 | 1,108.89 | |
| ATTNModel=Pythia-6.9B2025.10 | 614.1 | 1,115.6 |