TODO

Can LLMs use print debugging

Can you steer OCR with paliGemma using colPali (steering vectors)

OCR pipeline

VHS to ebay

Image to image fast demo