{"id":4983,"date":"2024-06-17T04:16:00","date_gmt":"2024-06-16T19:16:00","guid":{"rendered":"https:\/\/devneko.jp\/wordpress\/?p=4983"},"modified":"2024-06-17T04:16:00","modified_gmt":"2024-06-16T19:16:00","slug":"nemotron-4-340b","status":"publish","type":"post","link":"https:\/\/devneko.jp\/wordpress\/?p=4983","title":{"rendered":"Nemotron-4 340B"},"content":{"rendered":"\n<p>NVIDIA\u304b\u3089\u30aa\u30fc\u30d7\u30f3\u306a\u30e2\u30c7\u30ebNemotron-4 340B\u304c\u767a\u8868\u3055\u308c\u305f\u3002<\/p>\n\n\n\n<p><a href=\"https:\/\/blogs.nvidia.com\/blog\/nemotron-4-synthetic-data-generation-llm-training\/\">NVIDIA Releases Open Synthetic Data Generation Pipeline for Training Large Language Models | NVIDIA Blog<\/a><\/p>\n\n\n\n<pre class=\"wp-block-preformatted\">NVIDIA today announced Nemotron-4 340B, a family of open models that developers can use to generate synthetic data for training large language models (LLMs) for commercial applications across healthcare, finance, manufacturing, retail and every other industry.<\/pre>\n\n\n\n<p>\u3068\u306e\u3053\u3068\u3067Synthetic data\u306e\u751f\u6210\u3092\u76ee\u7684\u3068\u3057\u3066\u6319\u3052\u3066\u3044\u308b\u73cd\u3057\u3044\u30bf\u30a4\u30d7\u306e\u30e2\u30c7\u30eb\u3002\u30e9\u30a4\u30bb\u30f3\u30b9\u3082\u5bdb\u5bb9\uff08<a href=\"https:\/\/developer.download.nvidia.com\/licenses\/nvidia-open-model-license-agreement-june-2024.pdf\">nvidia-open-model-license-agreement-june-2024.pdf<\/a>\uff09\u3067<\/p>\n\n\n\n<pre class=\"wp-block-preformatted\">\u2022 Models are commercially useable.<br>\u2022 You are free to create and distribute Derivative Models.<br>\u2022 NVIDIA does not claim ownership to any outputs generated using the Models or Model Derivatives.<\/pre>\n\n\n\n<p>\u3068\u306e\u3053\u3068\u3002\u4e0b\u8a18\u6761\u9805\u3082\u7279\u5fb4\u7684\u3002Apache-2\u30e9\u30a4\u30bb\u30f3\u30b9\u306e\u7279\u8a31\u6761\u9805\u306b\u8fd1\u3044\u3082\u306e\u3092\u611f\u3058\u308b\u3002<\/p>\n\n\n\n<pre class=\"wp-block-preformatted\">If You institute copyright or patent litigation against any entity (including a crossclaim or counterclaim in a lawsuit) alleging that the Model or a Derivative Model constitutes direct or contributory copyright or patent infringement, then any licenses granted to You under this Agreement for that Model or Derivative Model will terminate as of the date such litigation is filed.\t<\/pre>\n\n\n\n<p>\u6027\u80fd\u306f\u9ad8\u304fllama3 70B\u3092\u8d85\u3048\u3066\u3044\u305d\u3046\u3002\u307e\u305f\u3001Nemotron-4-340B-Reward\u306f<a href=\"https:\/\/github.com\/allenai\/reward-bench\">GitHub &#8211; allenai\/reward-bench: RewardBench: the first evaluation tool for reward models.<\/a>\u3067\u5546\u7528\u30e2\u30c7\u30eb\uff08GPT-4o\u3084Gemini Pro\u306a\u3069\uff09\u3092\u4e0a\u56de\u308b\u3002<\/p>\n\n\n\n<p>fine tuning\u3092\u542b\u3081\u30ed\u30fc\u30ab\u30ebLLM\u3092\u4f5c\u308d\u3046\u3068\u8003\u3048\u308b\u3068\u304d\u306b\u975e\u5e38\u306b\u6709\u7528\u306a\u30e2\u30c7\u30eb\u3067\u30cf\u30fc\u30c9\u30a6\u30a7\u30a2\u3092\u62bc\u3055\u3048\u3066\u3044\u308bNVIDIA\u3089\u3057\u3044\u52d5\u304d\u3002<\/p>\n\n\n\n<p>Reward\u30e2\u30c7\u30eb\u306b\u3064\u3044\u3066\u306f\u4e0b\u8a18\u8ad6\u6587\u3082\u53c2\u8003\u306b\u306a\u308b\u3002<\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li><strong>HelpSteer2: Open-source dataset for training top-performing reward models\u00a0<\/strong>[9.2]<br>\u6211\u3005\u306f\u30d1\u30fc\u30df\u30c3\u30b7\u30d6\u306b\u30e9\u30a4\u30bb\u30f3\u30b9\u3055\u308c\u305f\u9078\u597d\u30c7\u30fc\u30bf\u30bb\u30c3\u30c8\u3067\u3042\u308bHelpSteer2\u3092\u958b\u767a\u3057\u305f\u3002 HelpSteer2\u306f1\u4e07\u306e\u30ec\u30b9\u30dd\u30f3\u30b9\u30da\u30a2\u3067\u69cb\u6210\u3055\u308c\u3066\u3044\u308b\u3002 \u672c\u7a3f\u3067\u306f,\u5831\u5968\u30e2\u30c7\u30eb\u306b\u3088\u3063\u3066\u4e88\u6e2c\u3055\u308c\u308b\u591a\u5c5e\u6027\u30b9\u30b3\u30a2\u3092\u52b9\u679c\u7684\u306b\u6d3b\u7528\u3067\u304d\u308b\u30e2\u30c7\u30eb\u30a2\u30e9\u30a4\u30e1\u30f3\u30c8\u624b\u6cd5\u3067\u3042\u308bSteerLM 2.0\u3092\u63d0\u6848\u3059\u308b\u3002<br><a href=\"http:\/\/arxiv.org\/abs\/2406.08673v1\">\u8ad6\u6587<\/a>\u00a0\u00a0<a href=\"https:\/\/fugumt.com\/fugumt\/paper_check\/2406.08673v1\">\u53c2\u8003\u8a33\uff08\u30e1\u30bf\u30c7\u30fc\u30bf\uff09<\/a>\u00a0 \u00a0(Wed, 12 Jun 2024 22:28:08 GMT)<\/li>\n\n\n\n<li>NVIDIA\u306b\u3088\u308bReward\u30e2\u30c7\u30eb\u7528\u30c7\u30fc\u30bf\u30bb\u30c3\u30c8\u3068\u624b\u6cd5\u306e\u63d0\u6848<\/li>\n\n\n\n<li>\u30c7\u30fc\u30bf\u306f<a href=\"https:\/\/huggingface.co\/datasets\/nvidia\/HelpSteer2\">nvidia\/HelpSteer2 \u00b7 Datasets at Hugging Face<\/a>\u3000\u30ea\u30dd\u30b8\u30c8\u30ea\u306f<a href=\"https:\/\/github.com\/NVIDIA\/NeMo-Aligner\">GitHub &#8211; NVIDIA\/NeMo-Aligner: Scalable toolkit for efficient model alignment<\/a><\/li>\n<\/ul>\n","protected":false},"excerpt":{"rendered":"<p>NVIDIA\u304b\u3089\u30aa\u30fc\u30d7\u30f3\u306a\u30e2\u30c7\u30ebNemotron-4 340B\u304c\u767a\u8868\u3055\u308c\u305f\u3002 NVIDIA Releases Open Synthetic Data Generation Pipeline for Training La &hellip; <a href=\"https:\/\/devneko.jp\/wordpress\/?p=4983\" class=\"more-link\"><span class=\"screen-reader-text\">&#8220;Nemotron-4 340B&#8221; \u306e<\/span>\u7d9a\u304d\u3092\u8aad\u3080<\/a><\/p>\n","protected":false},"author":1,"featured_media":0,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[2],"tags":[223,293,390],"class_list":["post-4983","post","type-post","status-publish","format-standard","hentry","category-arxiv","tag-llm","tag-oss","tag-synthetic-data"],"_links":{"self":[{"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=\/wp\/v2\/posts\/4983","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=4983"}],"version-history":[{"count":0,"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=\/wp\/v2\/posts\/4983\/revisions"}],"wp:attachment":[{"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=%2Fwp%2Fv2%2Fmedia&parent=4983"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=%2Fwp%2Fv2%2Fcategories&post=4983"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=%2Fwp%2Fv2%2Ftags&post=4983"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}