{"author":"Curiositry","children":[{"author":"gumboshoes","children":[],"created_at":"2026-02-07T00:43:39.000Z","created_at_i":1770425019,"id":46920122,"options":[],"parent_id":46920056,"points":null,"story_id":46920056,"text":"As is too often the case, it doesn&#x27;t follow what I consider to be a fundamental OCR rule: the input document type should also be a possible output document type. So, PDF in should mean that composited PDF output is possible. This is necessary for groundtruthing.","title":null,"type":"comment","url":null},{"author":"jauntywundrkind","children":[],"created_at":"2026-02-07T04:51:47.000Z","created_at_i":1770439907,"id":46921412,"options":[],"parent_id":46920056,"points":null,"story_id":46920056,"text":"Also at <a href=\"https:&#x2F;&#x2F;huggingface.co&#x2F;datalab-to&#x2F;chandra\" rel=\"nofollow\">https:&#x2F;&#x2F;huggingface.co&#x2F;datalab-to&#x2F;chandra</a>","title":null,"type":"comment","url":null}],"created_at":"2026-02-07T00:32:39.000Z","created_at_i":1770424359,"id":46920056,"options":[],"parent_id":null,"points":6,"story_id":46920056,"text":null,"title":"Chandra-OCR","type":"story","url":"https://github.com/datalab-to/chandra"}