Open milenamilka755 opened 1 month ago
From a few tests, it seems like the models are generally bad at writing non-ascii or uncommon characters, including e.g. these Polish diacritical marks, German Umlauts, French accents, Greek letters, and mathematical symbols
I know that models cope badly (not sure why) with non-ascii or uncommon characters. The question is when there will finally be a model that can handle it. We have been waiting for this for several years. Everyone marvels at the capabilities of various models for generating graphics, and they can't handle simple lettering. Why is it so difficult?
NO model can generate Polish diacritical marks ĄĘŁŃÓŚŻŹ ąęłńóśżź