Human Judgment Correlation On Flickr8K Cf
Metrics
Kendall's Tau-b
Results
Performance results of various models on this benchmark
Comparison Table
Model Name | Kendall's Tau-b |
---|---|
clipscore-a-reference-free-evaluation-metric | 34.4 |
mutual-information-divergence-a-unified | 37.3 |
clipscore-a-reference-free-evaluation-metric | 36.4 |