FactScore
Benchmarks
| Task Name | Dataset Name | SOTA Result | Trend | |
|---|---|---|---|---|
| Long-form Factuality | FActScore | FActScore59.3 | 32 | |
| Factuality Evaluation | FActScore | Pairwise Score84.5 | 24 | |
| Honesty Evaluation | FActScore v1.0 | Score47.3 | 20 | |
| Long-form text generation | FactScore | FactScore68.1 | 20 | |
| Claim-level Uncertainty Quantification | FactScore English (test) | ROC-AUC71 | 20 | |
| Fact-checking of atomic claims | FactScore English | PR-AUC0.34 | 20 | |
| Knowledge Graph Factuality Evaluation | FActScore | FActScore84 | 16 | |
| Long-form Factuality Verification | FactScore | Precision@165.41 | 15 | |
| Factual Text Generation | FactScore | AURC0.7345 | 14 | |
| long-form generation | FActScore (test) | AUROC0.8581 | 12 | |
| Factuality Generation | FActScore (test) | Number of Facts20.4 | 12 | |
| Factuality Evaluation | FactScore (unlabeled) | US (%)76.4 | 10 | |
| Factuality Evaluation | FactScore (labeled) | LS Score (%)64.8 | 10 | |
| Factuality and Hallucination Detection | FactScore | FactScore96.1 | 9 | |
| Long-form Factuality Calibration | FactScore | ECE0.076 | 8 | |
| Consistency Assessment of Generated Reference Points | FactScore LLM-based evaluation | Score86.36 | 6 | |
| Long-form generation | FactScore 200-word response length constraint | Response Coverage (%)98.9 | 5 | |
| Knowledge Graph Factuality Evaluation | FActScore Context and General truth | FActScore80.2 | 2 | |
| Knowledge Graph Factuality Evaluation | FActScore* Context only | FActScore76.9 | 2 |
Showing 19 of 19 rows