{"id":7841,"date":"2025-12-02T04:33:00","date_gmt":"2025-12-01T19:33:00","guid":{"rendered":"https:\/\/devneko.jp\/wordpress\/?p=7841"},"modified":"2025-11-29T17:01:25","modified_gmt":"2025-11-29T08:01:25","slug":"nvidia-nemotron-parse-1-1-nemotron-flash","status":"publish","type":"post","link":"https:\/\/devneko.jp\/wordpress\/?p=7841","title":{"rendered":"NVIDIA Nemotron Parse 1.1 \/ Nemotron-Flash"},"content":{"rendered":"\n<ul class=\"wp-block-list\">\n<li><strong>NVIDIA Nemotron Parse 1.1\u00a0<\/strong>[52.6]<br>Nemotron-Parse-1.1\u306f\u8efd\u91cf\u306a\u6587\u66f8\u89e3\u6790\u3068OCR\u30e2\u30c7\u30eb\u3067\u3042\u308b\u3002 \u4e00\u822c\u7684\u306aOCR\u3001\u30de\u30fc\u30af\u30c0\u30a6\u30f3\u30d5\u30a9\u30fc\u30de\u30c3\u30c8\u3001\u69cb\u9020\u5316\u30c6\u30fc\u30d6\u30eb\u89e3\u6790\u3001\u753b\u50cf\u3001\u30c1\u30e3\u30fc\u30c8\u3001\u30c0\u30a4\u30a2\u30b0\u30e9\u30e0\u304b\u3089\u306e\u30c6\u30ad\u30b9\u30c8\u62bd\u51fa\u306a\u3069\u3001\u6539\u5584\u3055\u308c\u305f\u6a5f\u80fd\u3092\u63d0\u4f9b\u3059\u308b\u3002 \u6211\u3005\u306f\u3001\u3088\u308a\u5e83\u7bc4\u306aNemotron-VLM-v2\u30c7\u30fc\u30bf\u30bb\u30c3\u30c8\u306e\u4e00\u90e8\u3068\u3057\u3066\u3001\u30c8\u30ec\u30fc\u30cb\u30f3\u30b0\u30c7\u30fc\u30bf\u306e\u30b5\u30d6\u30bb\u30c3\u30c8\u3068\u3068\u3082\u306b\u3001Huggingface\u306e\u30e2\u30c7\u30eb\u30a6\u30a7\u30a4\u30c8\u3068\u6700\u9069\u5316\u3055\u308c\u305fNIM\u30b3\u30f3\u30c6\u30ca\u3092\u516c\u958b\u3057\u3066\u3044\u307e\u3059\u3002<br><a href=\"http:\/\/arxiv.org\/abs\/2511.20478v1\">\u8ad6\u6587<\/a>\u00a0\u00a0<a href=\"https:\/\/fugumt.com\/fugumt\/paper_check\/2511.20478v1\">\u53c2\u8003\u8a33\uff08\u30e1\u30bf\u30c7\u30fc\u30bf\uff09<\/a>\u00a0 \u00a0(Tue, 25 Nov 2025 16:41:25 GMT)<\/li>\n\n\n\n<li>\u300cNemotron-Parse-1.1 follows an encoder-decoder architecture with 885M parameters, including a compact 256M-parameter language decoder.\u300d\u3068\u3044\u3046OCR\u95a2\u9023\u30e2\u30c7\u30eb\u3002\uff08\u30bf\u30b9\u30af\u306b\u30d5\u30a3\u30c3\u30c8\u3057\u3066\u3044\u308b\u3068\u3044\u3046\u3053\u3068\u3082\u3042\u308b\u306e\u3060\u308d\u3046\u304c\uff09decoder only\u3067\u306f\u306a\u3044\u3002<\/li>\n\n\n\n<li>\u30ea\u30dd\u30b8\u30c8\u30ea\u306f<a href=\"https:\/\/huggingface.co\/nvidia\/NVIDIA-Nemotron-Parse-v1.1-TC\">nvidia\/NVIDIA-Nemotron-Parse-v1.1-TC \u00b7 Hugging Face<\/a><\/li>\n<\/ul>\n\n\n\n<ul class=\"wp-block-list\">\n<li><strong>Nemotron-Flash: Towards Latency-Optimal Hybrid Small Language Models\u00a0<\/strong>[97.6]<br>\u672c\u7814\u7a76\u306e\u76ee\u7684\u306f\u3001SLM\u306e\u30ea\u30a2\u30eb\u30bf\u30a4\u30e0\u30ec\u30a4\u30c6\u30f3\u30b7\u306e\u4e3b\u8981\u306a\u6c7a\u5b9a\u8981\u56e0\u3092\u7279\u5b9a\u3057\u3001SLM\u306e\u8a2d\u8a08\u3068\u30c8\u30ec\u30fc\u30cb\u30f3\u30b0\u306e\u305f\u3081\u306e\u4e00\u822c\u5316\u53ef\u80fd\u306a\u539f\u5247\u3068\u65b9\u6cd5\u8ad6\u3092\u63d0\u4f9b\u3059\u308b\u3053\u3068\u3067\u3042\u308b\u3002 \u6211\u3005\u306fNemotron-Flash\u3068\u547c\u3070\u308c\u308b\u30cf\u30a4\u30d6\u30ea\u30c3\u30c9SLM\u306e\u65b0\u305f\u306a\u30d5\u30a1\u30df\u30ea\u30fc\u3092\u5c0e\u5165\u3057\u3001\u6700\u5148\u7aefSLM\u306e\u7cbe\u5ea6\u30fb\u52b9\u7387\u306e\u30d5\u30ed\u30f3\u30c6\u30a3\u30a2\u3092\u5927\u5e45\u306b\u5411\u4e0a\u3055\u305b\u308b\u3002<br><a href=\"http:\/\/arxiv.org\/abs\/2511.18890v1\">\u8ad6\u6587<\/a>\u00a0\u00a0<a href=\"https:\/\/fugumt.com\/fugumt\/paper_check\/2511.18890v1\">\u53c2\u8003\u8a33\uff08\u30e1\u30bf\u30c7\u30fc\u30bf\uff09<\/a>\u00a0 \u00a0(Mon, 24 Nov 2025 08:46:36 GMT)<\/li>\n\n\n\n<li>\u300cRather than merely offering a smaller LLM, this work re-imagines small models from the perspective of real- world latency and throughput, systematically explor- ing the key architectural and training factors essential for developing latency-optimal SLMs. By analyzing optimal depth\u2013width ratios, strategically combining efficient attention operators through an evolutionary search framework, and enhancing training with weight normalization and meta tokens, we establish a comprehensive framework that significantly improves both real-device latency and accuracy, and deliver the Nemotron-Flash model family that advances the SOTA accuracy\u2013latency frontier.\u300d\u3068\u30a2\u30fc\u30ad\u30c6\u30af\u30c1\u30e3\u8a2d\u8a08\u306b\u8e0f\u307f\u8fbc\u3093\u3067\u306eSLM\u306e\u63a2\u6c42<\/li>\n\n\n\n<li>\u30ea\u30dd\u30b8\u30c8\u30ea\u306f<a href=\"https:\/\/huggingface.co\/nvidia\/Nemotron-Flash-3B\">nvidia\/Nemotron-Flash-3B \u00b7 Hugging Face<\/a><\/li>\n<\/ul>\n","protected":false},"excerpt":{"rendered":"","protected":false},"author":1,"featured_media":0,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[2],"tags":[117,365],"class_list":["post-7841","post","type-post","status-publish","format-standard","hentry","category-arxiv","tag-document-understanding","tag-slm"],"_links":{"self":[{"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=\/wp\/v2\/posts\/7841","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=%2Fwp%2Fv2%2Fcomments&post=7841"}],"version-history":[{"count":1,"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=\/wp\/v2\/posts\/7841\/revisions"}],"predecessor-version":[{"id":7842,"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=\/wp\/v2\/posts\/7841\/revisions\/7842"}],"wp:attachment":[{"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=%2Fwp%2Fv2%2Fmedia&parent=7841"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=%2Fwp%2Fv2%2Fcategories&post=7841"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=%2Fwp%2Fv2%2Ftags&post=7841"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}