goodcatch Open-source attempt to implement tiny vision-language model which works well with text-rich images