In most cases you should have the LLM write a program to process the files instead of feeding them to the LLM itself.
And a program which parses handwritten checks and invoices which range from clean/crisp printouts through multiple-generation photocopies/faxes, or even a photograph of a folded printout would be?
And a program which parses handwritten checks and invoices which range from clean/crisp printouts through multiple-generation photocopies/faxes, or even a photograph of a folded printout would be?