Text-to-Chart Retrieval on VisText L2+L3 Caption benchmark leaderboard · SOTA2 Research

Text-to-Chart Retrieval on VisText L2+L3 Caption

0.8458R@5ChartFinder

Evaluation Results

Method	Links
ChartFinder2025.05		0.8458	0.8912	0.752	0.7878
Long-CLIP-L2025.05		0.8299	0.8866	0.7343	0.7709
ChartFinder-B2025.05		0.7948	0.8401	0.6964	0.7311
Long-CLIP-B2025.05		0.7812	0.8345	0.6799	0.7173
Jina-CLIP-v22025.05		0.7449	0.805	0.6272	0.6701
CLIP2025.05		0.6825	0.7483	0.5681	0.6114
CLIP-DPR2025.05		0.6474	0.7211	0.5351	0.5797
VISTA2025.05		0.6395	0.7086	0.3761	0.418
CoCa2025.05		0.5896	0.6644	0.5054	0.543
Univl-DR2025.05		0.5283	0.6077	0.4261	0.4693
EVA-CLIP2025.05		0.4943	0.5726	0.3939	0.4364
MARVEL2025.05		0.398	0.4864	0.3125	0.3537