Md Dilshadur Rahman

AnnoBench reveals that the right combination of input representation and prompt specificity can significantly enhance the quality of visualization annotations, challenging existing assumptions in the field.

Accessible context can shift MLLMs toward more reliable claims, but images alone may not enhance interpretative accuracy.