MuST-Bench
Benchmarks
TaskDatasetSOTA resultTrendResults
MuST-Bench
3.89GPT-4V Fidelity (EN)
4
MuST-Bench Korean
40.69Style Fidelity
4
MuST-Bench Chinese
0.404Style Fidelity
4
MuST-Bench English
0.405Style Fidelity
4
MuST-Bench Korean
65.77OCR Accuracy
4
MuST-Bench Chinese
74.81OCR Accuracy
4
MuST-Bench English
71.63OCR Accuracy
4