CLIP-S

Reported on 3 benchmarks across 2 tasks · 1 paper

Note: results are matched by exact model name. Different papers may use the same name for different model variants.

Reasoning3 results

Human Judgment CorrelationonFlickr8k-CF
Kendall's Tau-b· 2021-04-18
34.4
best: 37.3 (MID)
CLIPScore: A Reference-free Evaluation Metric for Image Captioning arXiv:2104.08718
Human Judgment CorrelationonFlickr8k-Expert
Kendall's Tau-c· 2021-04-18
51.2
best: 54.9 (MID)
CLIPScore: A Reference-free Evaluation Metric for Image Captioning arXiv:2104.08718
Human Judgment ClassificationonPascal-50S
Mean Accuracy· 2021-04-18
80.7
best: 85.2 (MID)
CLIPScore: A Reference-free Evaluation Metric for Image Captioning arXiv:2104.08718