Factual Precision Evaluation on Geographic
86.7FactScoreMistralINST
Evaluation Results
| Method | Links | |
|---|---|---|
| MistralINSTScenario=INFO, CORE=w/o2024.07 | 86.7 | |
| MistralINSTScenario=REP, CORE=w/o2024.07 | 84.9 | |
| GPT-2Scenario=REP, CORE=w/o2024.07 | 80.7 | |
| MistralINSTScenario=NORMAL, CORE=w/o2024.07 | 79.9 | |
| MistralINSTScenario=NORMAL, CORE=w/2024.07 | 78.3 | |
| GPT-2Scenario=INFO, CORE=w/o2024.07 | 75.8 | |
| MistralINSTScenario=INFO, CORE=w/2024.07 | 53.7 | |
| MistralINSTScenario=REP, CORE=w/2024.07 | 32.7 | |
| GPT-2Scenario=INFO, CORE=w/2024.07 | 1.68 | |
| GPT-2Scenario=REP, CORE=w/2024.07 | 1.22 |