{"id":168490,"date":"2023-06-06T16:00:00","date_gmt":"2023-06-06T15:00:00","guid":{"rendered":"https:\/\/liora.io\/en\/?p=168490"},"modified":"2026-08-09T18:23:28","modified_gmt":"2026-08-09T17:23:28","slug":"the-new-champion-of-open-source-llm-falcon","status":"publish","type":"post","link":"https:\/\/liora.io\/en\/the-new-champion-of-open-source-llm-falcon","title":{"rendered":"The new champion of open source LLM, Falcon"},"content":{"rendered":"\n<p><strong>In the open source community, LLaMA had the effect of a technological leap, giving independent developers access to a large GPT-level language model. Today, Abu Dhabi&#8217;s Institute of Innovation and Technology (IIT) unveils Falcon, an open source LLM that outperforms LLaMA.<\/strong><\/p>\n\n\n<h2 class=\"wp-block-heading\" id=\"what-is-falcon\">What is Falcon?<\/h2>\n\n\n<p>Falcon is presented as <a href=\"https:\/\/liora.io\/en\/large-language-models-llm-everything-you-need-to-know\" rel=\"noopener\" target=\"_blank\">the most powerful language model to date<\/a>, with three possible variants: <strong>Falcon 1B, 7B and 40B<\/strong>. Smaller than LLaMA, with <strong>40 billion parameters<\/strong> versus 65, it nevertheless outperforms the latter. According to <strong>Hugging Face&#8217;s<\/strong> evaluation criteria (IA2 Reasoning Challenge, HellaSwag, MMLU and TruthfulQA), Falcon 40B Instruct, a Falcon variant, and Falcon 40B are more powerful than LLaMA in terms of performance.<\/p>\n\n\n<p>This model is <strong>multilingual<\/strong>, understanding English, German, Spanish and French, and Dutch, Italian, Romanian, Portuguese, Czech, Polish and Swedish.<\/p>\n\n\n<p>To achieve this result, IIT used <strong>a dataset of 1,000 billion tokens<\/strong> and <strong>a pipeline<\/strong> capable of extracting verified content to ensure the quality of Falcon&#8217;s responses. This <a href=\"https:\/\/huggingface.co\/datasets\/tiiuae\/falcon-refinedweb\" rel=\"noopener\" target=\"_blank\">&#8220;refined-web&#8221;<\/a> dataset is also <strong>open source<\/strong>, so developers can train their IA to produce programs as powerful as, or even better than, those currently available.<\/p>\n\n\n<figure class=\"wp-block-image size-full\" style=\"margin-top:32px;margin-bottom:32px\"><img alt=\"Illustration for What is Falcon?\" decoding=\"async\" height=\"800\" loading=\"lazy\" sizes=\"(max-width: 800px) 100vw, 800px\" src=\"https:\/\/liora.io\/app\/uploads\/sites\/9\/2023\/06\/4944103-1024x1024.jpg\" srcset=\"https:\/\/liora.io\/app\/uploads\/sites\/9\/2023\/06\/4944103-1024x1024.jpg 1024w, https:\/\/liora.io\/app\/uploads\/sites\/9\/2023\/06\/4944103-300x300.jpg 300w, https:\/\/liora.io\/app\/uploads\/sites\/9\/2023\/06\/4944103-150x150.jpg 150w, https:\/\/liora.io\/app\/uploads\/sites\/9\/2023\/06\/4944103-768x768.jpg 768w, https:\/\/liora.io\/app\/uploads\/sites\/9\/2023\/06\/4944103-1536x1536.jpg 1536w, https:\/\/liora.io\/app\/uploads\/sites\/9\/2023\/06\/4944103.jpg 2000w\" style=\"width:100%;height:auto\" width=\"800\"\/><\/figure>\n\n\n<h2 class=\"wp-block-heading\" id=\"how-useful-will-it-be\">How useful will it be?<\/h2>\n\n\n<p>Unlike its predecessor, developers will be able to use <a href=\"https:\/\/huggingface.co\/spaces\/HuggingFaceH4\/open_llm_leaderboard\" rel=\"noopener\" target=\"_blank\">Falcon<\/a> for<strong> commercial purposes<\/strong>. Although LLaMA is open source, these weights remain private for Meta, which limits its commercialization. This is why Falcon&#8217;s models, which use <strong>a modified version of Apache 2.0<\/strong>, can suit the user&#8217;s needs.<\/p>\n\n\n<figure class=\"wp-block-image size-full\" style=\"margin-top:32px;margin-bottom:32px\"><img alt=\"Illustration for How useful will it be?\" decoding=\"async\" height=\"534\" loading=\"lazy\" sizes=\"(max-width: 800px) 100vw, 800px\" src=\"https:\/\/liora.io\/app\/uploads\/sites\/9\/2023\/06\/kaleidico-3V8xo5Gbusk-unsplash-1024x683.jpg\" srcset=\"https:\/\/liora.io\/app\/uploads\/sites\/9\/2023\/06\/kaleidico-3V8xo5Gbusk-unsplash-1024x683.jpg 1024w, https:\/\/liora.io\/app\/uploads\/sites\/9\/2023\/06\/kaleidico-3V8xo5Gbusk-unsplash-300x200.jpg 300w, https:\/\/liora.io\/app\/uploads\/sites\/9\/2023\/06\/kaleidico-3V8xo5Gbusk-unsplash-768x512.jpg 768w, https:\/\/liora.io\/app\/uploads\/sites\/9\/2023\/06\/kaleidico-3V8xo5Gbusk-unsplash-1536x1024.jpg 1536w, https:\/\/liora.io\/app\/uploads\/sites\/9\/2023\/06\/kaleidico-3V8xo5Gbusk-unsplash.jpg 1920w\" style=\"width:100%;height:auto\" width=\"800\"\/><\/figure>\n\n\n<p><strong>Developers trained<\/strong> to design new artificial intelligence will then be able to use Falcon to create a generation of <strong>even more powerful AIs<\/strong>. That&#8217;s why, if you&#8217;ve enjoyed this article and are considering a career in Data Science, don&#8217;t hesitate to check out <a href=\"https:\/\/liora.io\/en\/blog-en\" rel=\"noopener\" target=\"_blank\">our articles<\/a> or <a href=\"\/en\/courses\/data-ai\/\" rel=\"noopener\" target=\"_blank\">our training offers<\/a> on Liora.<\/p>\n\n\n<p><i>Source : huggingface.co<\/i><\/p>\n\n","protected":false},"excerpt":{"rendered":"<p>In the open source community, LLaMA had the effect of a technological leap, giving independent developers access to a large GPT-level language model. Today, Abu Dhabi&#8217;s Institute of Innovation and Technology (IIT) unveils Falcon, an open source LLM that outperforms LLaMA. What is Falcon? Falcon is presented as the most powerful language model to date, [&hellip;]<\/p>\n","protected":false},"author":74,"featured_media":168491,"comment_status":"open","ping_status":"open","sticky":false,"template":"elementor_theme","format":"standard","meta":{"_acf_changed":false,"editor_notices":[],"footnotes":""},"categories":[2433],"class_list":["post-168490","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-data-ai"],"acf":[],"_links":{"self":[{"href":"https:\/\/liora.io\/en\/wp-json\/wp\/v2\/posts\/168490","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/liora.io\/en\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/liora.io\/en\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/liora.io\/en\/wp-json\/wp\/v2\/users\/74"}],"replies":[{"embeddable":true,"href":"https:\/\/liora.io\/en\/wp-json\/wp\/v2\/comments?post=168490"}],"version-history":[{"count":2,"href":"https:\/\/liora.io\/en\/wp-json\/wp\/v2\/posts\/168490\/revisions"}],"predecessor-version":[{"id":210798,"href":"https:\/\/liora.io\/en\/wp-json\/wp\/v2\/posts\/168490\/revisions\/210798"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/liora.io\/en\/wp-json\/wp\/v2\/media\/168491"}],"wp:attachment":[{"href":"https:\/\/liora.io\/en\/wp-json\/wp\/v2\/media?parent=168490"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/liora.io\/en\/wp-json\/wp\/v2\/categories?post=168490"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}