Correlation with human judgments on Composite (test)
57.6Kendall's Tau-cPolos
Evaluation Results
| Method | Links | |
|---|---|---|
| Poloscategory=Learning-based metrics2024.02 | 57.6 | |
| RefPAC-Scategory=Learning-based metrics2024.02 | 57.3 | |
| UMICcategory=Learning-based metrics2024.02 | 56.1 | |
| MIDcategory=Similarity-based metrics2024.02 | 55.7 | |
| PAC-Scategory=Learning-based metrics2024.02 | 55.7 | |
| RefCLIP-Scategory=Similarity-based metrics2024.02 | 55.4 | |
| CLIP-Scategory=Similarity-based metrics2024.02 | 53.8 | |
| ViLBERTScorecategory=Similarity-based metrics2024.02 | 52.4 | |
| TIGERcategory=Similarity-based metrics2024.02 | 45.4 | |
| BARTScorecategory=Similarity-based metrics2024.02 | 43.5 | |
| SPARCScategory=Classic metrics2024.02 | 43.1 | |
| SPICEcategory=Classic metrics2024.02 | 40.3 | |
| METEORcategory=Classic metrics2024.02 | 38.9 | |
| CIDErcategory=Classic metrics2024.02 | 37.7 | |
| ROUGEcategory=Classic metrics2024.02 | 32.4 | |
| BLEUcategory=Classic metrics2024.02 | 30.6 | |
| MoverScorecategory=Similarity-based metrics2024.02 | 30.1 | |
| BERTScorecategory=Similarity-based metrics2024.02 | 30.1 |