{"about":{"site":"https://codewithpapers.app","non_affiliation":"Code with Papers and Syntology are not affiliated with, endorsed by, or sponsored by Papers with Code, Meta, or the pwc-archive mirror.","licence":"CC BY-SA 4.0","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","attribution":"https://codewithpapers.app/attribution","modified":"archive material modified by Syntology; see the attribution page"},"url":"/paper/teaching-machines-to-code-neural-markup","title":"Teaching Machines to Code: Neural Markup Generation with Visual Attention","arxiv_id":"1802.05415","date":"2018-02-15","proceeding":null,"authors":["Sumeet S. Singh"],"abstract":"We present a neural transducer model with visual attention that learns to\ngenerate LaTeX markup of a real-world math formula given its image. Applying\nsequence modeling and transduction techniques that have been very successful\nacross modalities such as natural language, image, handwriting, speech and\naudio; we construct an image-to-markup model that learns to produce\nsyntactically and semantically correct LaTeX markup code over 150 words long\nand achieves a BLEU score of 89%; improving upon the previous state-of-art for\nthe Im2Latex problem. We also demonstrate with heat-map visualization how\nattention helps in interpreting the model and can pinpoint (detect and\nlocalize) symbols on the image accurately despite having been trained without\nany bounding box data.","url_abs":"http://arxiv.org/abs/1802.05415v2","url_pdf":"http://arxiv.org/pdf/1802.05415v2.pdf","source":{"archive":"pwc-archive (Hugging Face), CC BY-SA 4.0","snapshot":"2025-07-28","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","row_kind":"abstracts"},"code_links":[{"paper_slug":"teaching-machines-to-code-neural-markup","repo_url":"https://github.com/untrix/im2latex","is_official":0,"mentioned_in_paper":1,"mentioned_in_github":1,"framework":"tf","reach":{"status":"ok","spdx":"AGPL-3.0"}}],"tasks":[{"task_slug":"math","task_name":"Math"},{"task_slug":"optical-character-recognition","task_name":"Optical Character Recognition (OCR)"}],"methods":[],"datasets_introduced":[{"slug":"i2l-140k","name":"I2L-140K","full_name":""},{"slug":"im2latex-90k","name":"Im2latex-90k","full_name":""}],"methods_introduced":[],"results":[{"leaderboard":"/sota/optical-character-recognition-on-i2l-140k","task":"Optical Character Recognition (OCR)","dataset":"I2L-140K","model":"I2L-NOPOOL","rank_in_archive_order":1,"of":2,"metrics":{"BLEU":"89.09%"},"uses_additional_data":false},{"leaderboard":"/sota/optical-character-recognition-on-i2l-140k","task":"Optical Character Recognition (OCR)","dataset":"I2L-140K","model":"I2L-STRIPS","rank_in_archive_order":2,"of":2,"metrics":{"BLEU":"89%"},"uses_additional_data":false},{"leaderboard":"/sota/optical-character-recognition-on-im2latex-1","task":"Optical Character Recognition (OCR)","dataset":"im2latex-100k","model":"I2L-STRIPS","rank_in_archive_order":1,"of":1,"metrics":{"BLEU":"88.86%"},"uses_additional_data":false}],"syntology":{"atlas_url":null,"mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}