3 ms·
VLM can already process both the document images and the query to produce an answer directly. Do we still need the intermediate OCR step?
by mingtianzhang 11mo ago
VLM can already process both the document images and the query to produce an answer directly. Do we still need the intermediate OCR step?