Factual Consistency on FACTOR (WIKI/NEWS Metrics)
64.43Factual Consistency (Wiki)HICD
Evaluation results
| Method | Links | ||
|---|---|---|---|
| HICDBackbone=LLaMA-3-8B-Instruct2025.03 | 64.43 | 75.5 | |
| HICDBackbone=Llama3-8B-Instruct2025.03 | 64.43 | 75.5 | |
| DeCoReBackbone=Llama3-8B-Instruct2025.03 | 62.33 | 75.45 | |
| HICDBackbone=Mistral-7b-v0.32025.03 | 61.82 | 76.51 | |
| HICDBackbone=Mistral-7B-v0.32025.03 | 61.8 | 76.58 | |
| DeCoReBackbone=Mistral-7b-v0.32025.03 | 61.32 | 76.7 | |
| ITIBackbone=Llama3-8B-Instruct2025.03 | 61.06 | 72.59 | |
| LLaMA-3-8B-InstructBackbone=LLaMA-3-8B-Instruct2025.03 | 61.02 | 74.51 | |
| Llama3-8B-InstructBackbone=Llama3-8B-Instruct2025.03 | 61.02 | 74.51 | |
| ITIBackbone=Mistral-7b-v0.32025.03 | 60.85 | 73.34 | |
| HICDBackbone=Qwen-7B2025.03 | 60.21 | 70.15 | |
| Mistral-7B-v0.3Backbone=Mistral-7B-v0.32025.03 | 59.31 | 74.22 | |
| Mistral-7b-v0.3Backbone=Mistral-7b-v0.32025.03 | 59.31 | 74.22 | |
| Qwen-7BBackbone=Qwen-7B2025.03 | 58.78 | 69.88 |