Faithfulness Evaluation on Fakepedia Double-Conflicting long-context
1,690.2AOPC ComponentFA
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| FAModel=Qwen2.5-7B2025.10 | 1,690.2 | 2,294.96 | |
| IGModel=Qwen2.5-7B2025.10 | 1,688.5 | 2,283.39 | |
| IGModel=Pythia-6.9B2025.10 | 1,685.4 | 2,499.73 | |
| FAModel=Pythia-6.9B2025.10 | 1,683.5 | 2,482.1 | |
| ATTNModel=Qwen2.5-7B2025.10 | 1,645.8 | 2,340.83 | |
| MechLightModel=Qwen2.5-7B2025.10 | 1,621.8 | 2,283.02 | |
| MechLightModel=Pythia-6.9B2025.10 | 1,023.4 | 2,500.7 | |
| ATTNModel=Pythia-6.9B2025.10 | 817.8 | 2,493.53 |