[untitled]1 points·by mingtianzhang·8 months ago·0 comments1 commentsPost comment[–]mingtianzhang·8 months agoreplyVLM can already process both the document images and the query to produce an answer directly. Do we still need the intermediate OCR step?