“vision-based” large language models. Odd restriction. Why not investigate text-based ones? Or is “vision-based” a technical term that encompasses models that

“vision-based” large language models. Odd restriction. Why not investigate text-based ones? Or is “vision-based” a technical term that encompasses models that were trained on text?