{"id":134949,"date":"2023-09-25T12:47:23","date_gmt":"2023-09-25T12:47:23","guid":{"rendered":"https:\/\/blog.finxter.com\/?p=1651783"},"modified":"2023-09-25T12:47:23","modified_gmt":"2023-09-25T12:47:23","slug":"5-best-open-source-llms-in-2023-two-minute-guide","status":"publish","type":"post","link":"https:\/\/sickgaming.net\/blog\/2023\/09\/25\/5-best-open-source-llms-in-2023-two-minute-guide\/","title":{"rendered":"5 Best Open-Source LLMs in 2023 (Two-Minute Guide)"},"content":{"rendered":"\n<div class=\"kk-star-ratings kksr-auto kksr-align-left kksr-valign-top\" data-payload='{&quot;align&quot;:&quot;left&quot;,&quot;id&quot;:&quot;1651783&quot;,&quot;slug&quot;:&quot;default&quot;,&quot;valign&quot;:&quot;top&quot;,&quot;ignore&quot;:&quot;&quot;,&quot;reference&quot;:&quot;auto&quot;,&quot;class&quot;:&quot;&quot;,&quot;count&quot;:&quot;1&quot;,&quot;legendonly&quot;:&quot;&quot;,&quot;readonly&quot;:&quot;&quot;,&quot;score&quot;:&quot;5&quot;,&quot;starsonly&quot;:&quot;&quot;,&quot;best&quot;:&quot;5&quot;,&quot;gap&quot;:&quot;5&quot;,&quot;greet&quot;:&quot;Rate this post&quot;,&quot;legend&quot;:&quot;5\\\/5 - (1 vote)&quot;,&quot;size&quot;:&quot;24&quot;,&quot;title&quot;:&quot;5 Best Open-Source LLMs in 2023 (Two-Minute Guide)&quot;,&quot;width&quot;:&quot;142.5&quot;,&quot;_legend&quot;:&quot;{score}\\\/{best} - ({count} {votes})&quot;,&quot;font_factor&quot;:&quot;1.25&quot;}'>\n<div class=\"kksr-stars\">\n<div class=\"kksr-stars-inactive\">\n<div class=\"kksr-star\" data-star=\"1\" style=\"padding-right: 5px\">\n<div class=\"kksr-icon\" style=\"width: 24px; height: 24px;\"><\/div>\n<\/p><\/div>\n<div class=\"kksr-star\" data-star=\"2\" style=\"padding-right: 5px\">\n<div class=\"kksr-icon\" style=\"width: 24px; height: 24px;\"><\/div>\n<\/p><\/div>\n<div class=\"kksr-star\" data-star=\"3\" style=\"padding-right: 5px\">\n<div class=\"kksr-icon\" style=\"width: 24px; height: 24px;\"><\/div>\n<\/p><\/div>\n<div class=\"kksr-star\" data-star=\"4\" style=\"padding-right: 5px\">\n<div class=\"kksr-icon\" style=\"width: 24px; height: 24px;\"><\/div>\n<\/p><\/div>\n<div class=\"kksr-star\" data-star=\"5\" style=\"padding-right: 5px\">\n<div class=\"kksr-icon\" style=\"width: 24px; height: 24px;\"><\/div>\n<\/p><\/div>\n<\/p><\/div>\n<div class=\"kksr-stars-active\" style=\"width: 142.5px;\">\n<div class=\"kksr-star\" style=\"padding-right: 5px\">\n<div class=\"kksr-icon\" style=\"width: 24px; height: 24px;\"><\/div>\n<\/p><\/div>\n<div class=\"kksr-star\" style=\"padding-right: 5px\">\n<div class=\"kksr-icon\" style=\"width: 24px; height: 24px;\"><\/div>\n<\/p><\/div>\n<div class=\"kksr-star\" style=\"padding-right: 5px\">\n<div class=\"kksr-icon\" style=\"width: 24px; height: 24px;\"><\/div>\n<\/p><\/div>\n<div class=\"kksr-star\" style=\"padding-right: 5px\">\n<div class=\"kksr-icon\" style=\"width: 24px; height: 24px;\"><\/div>\n<\/p><\/div>\n<div class=\"kksr-star\" style=\"padding-right: 5px\">\n<div class=\"kksr-icon\" style=\"width: 24px; height: 24px;\"><\/div>\n<\/p><\/div>\n<\/p><\/div>\n<\/div>\n<div class=\"kksr-legend\" style=\"font-size: 19.2px;\"> 5\/5 &#8211; (1 vote) <\/div>\n<\/p><\/div>\n<p>Open-source research on <a href=\"https:\/\/blog.finxter.com\/how-to-run-large-language-models-llms-in-your-command-line\/\">large language models (LLMs)<\/a> is crucial for democratizing this powerful technology. <\/p>\n<p>Although open-source LLMs are now widely used and studied, they faced initial challenges and criticism. Early attempts at creating open-source LLMs like <a href=\"https:\/\/arxiv.org\/abs\/2205.01068\">OPT<\/a> and <a href=\"https:\/\/bigscience.huggingface.co\/blog\/bloom\">BLOOM<\/a> had poor performance compared to closed-source models. <\/p>\n<p><strong>This led researchers to realize the need for higher-quality base models pre-trained on larger datasets with trillions (!) of tokens!<\/strong> <\/p>\n<ul>\n<li><strong>OPT<\/strong>: 180 billion tokens<\/li>\n<li><strong>BLOOM<\/strong>: 341 billion tokens<\/li>\n<li><strong>LLaMa<\/strong>: 1.4 trillion tokens<\/li>\n<li><strong>MPT<\/strong>: 1 trillion tokens<\/li>\n<li><strong>Falcon<\/strong>: 1.5 trillion tokens<\/li>\n<li><mark style=\"background-color:rgba(0, 0, 0, 0)\" class=\"has-inline-color has-accent-color\"><strong>LLaMA 2<\/strong>: 2 trillion tokens<\/mark><\/li>\n<\/ul>\n<p>However, pre-training these models is expensive and requires organizations with sufficient funding to make them freely available to the community.<\/p>\n<p>This article focuses on high-performing open-source base models significantly improving the field. A great graphic of the historic context of open-source LLMs is presented on the <a href=\"https:\/\/python.langchain.com\/docs\/guides\/local_llms\">Langchain page<\/a>:<\/p>\n<div class=\"wp-block-image\">\n<figure class=\"aligncenter size-large\"><img decoding=\"async\" fetchpriority=\"high\" width=\"1024\" height=\"480\" src=\"https:\/\/blog.finxter.com\/wp-content\/uploads\/2023\/09\/OSS_LLM_overview-b0a96cc35216ec43c3ccde7ed1140854-1024x480.png\" alt=\"\" class=\"wp-image-1651789\" srcset=\"https:\/\/blog.finxter.com\/wp-content\/uploads\/2023\/09\/OSS_LLM_overview-b0a96cc35216ec43c3ccde7ed1140854-1024x480.png 1024w, https:\/\/blog.finxter.com\/wp-content\/uploads\/2023\/09\/OSS_LLM_overview-b0a96cc35216ec43c3ccde7ed1140854-300x141.png 300w, https:\/\/blog.finxter.com\/wp-content\/uploads\/2023\/09\/OSS_LLM_overview-b0a96cc35216ec43c3ccde7ed1140854-768x360.png 768w, https:\/\/blog.finxter.com\/wp-content\/uploads\/2023\/09\/OSS_LLM_overview-b0a96cc35216ec43c3ccde7ed1140854-1536x720.png 1536w, https:\/\/blog.finxter.com\/wp-content\/uploads\/2023\/09\/OSS_LLM_overview-b0a96cc35216ec43c3ccde7ed1140854.png 1689w\" sizes=\"(max-width: 1024px) 100vw, 1024px\" \/><\/figure>\n<\/div>\n<p>How can we determine the best of those? Easy, with Chatbot leaderboards like <a href=\"https:\/\/huggingface.co\/spaces\/lmsys\/chatbot-arena-leaderboard\">this<\/a> on Hugginface:<\/p>\n<figure class=\"wp-block-image size-large\"><img loading=\"lazy\" decoding=\"async\" width=\"1024\" height=\"754\" src=\"https:\/\/blog.finxter.com\/wp-content\/uploads\/2023\/09\/image-118-1024x754.png\" alt=\"\" class=\"wp-image-1651790\" srcset=\"https:\/\/blog.finxter.com\/wp-content\/uploads\/2023\/09\/image-118-1024x754.png 1024w, https:\/\/blog.finxter.com\/wp-content\/uploads\/2023\/09\/image-118-300x221.png 300w, https:\/\/blog.finxter.com\/wp-content\/uploads\/2023\/09\/image-118-768x566.png 768w, https:\/\/blog.finxter.com\/wp-content\/uploads\/2023\/09\/image-118.png 1078w\" sizes=\"auto, (max-width: 1024px) 100vw, 1024px\" \/><\/figure>\n<p>At the time of writing, the best non-commercial LLM is <a href=\"https:\/\/blog.finxter.com\/gpt4all-vs-vicuna-battle-of-open-source-llms-%e2%9a%94%ef%b8%8f\/\">Vicuna-33B<\/a>. Of course, closed-source <a href=\"https:\/\/blog.finxter.com\/20-ways-to-make-money-with-gpt-4\/\">GPT-4<\/a> by OpenAI and Claude by Anthropic models are the best. <\/p>\n<p>By the way, feel free to check out my article on Claude-2 proven to be one of the most powerful free but closed-source LLMs:<\/p>\n<div class=\"wp-block-image\">\n<figure class=\"aligncenter size-large\"><a href=\"https:\/\/blog.finxter.com\/claude-2-read-ten-papers-in-one-prompt-with-massive-200k-token-context\/?tl_inbound=1&amp;tl_target_all=1&amp;tl_form_type=1&amp;tl_period_type=3\" target=\"_blank\" rel=\"noreferrer noopener\"><img loading=\"lazy\" decoding=\"async\" width=\"1024\" height=\"574\" src=\"https:\/\/blog.finxter.com\/wp-content\/uploads\/2023\/09\/image-157-2-1024x574.png\" alt=\"\" class=\"wp-image-1651791\" srcset=\"https:\/\/blog.finxter.com\/wp-content\/uploads\/2023\/09\/image-157-2-1024x574.png 1024w, https:\/\/blog.finxter.com\/wp-content\/uploads\/2023\/09\/image-157-2-300x168.png 300w, https:\/\/blog.finxter.com\/wp-content\/uploads\/2023\/09\/image-157-2-768x430.png 768w, https:\/\/blog.finxter.com\/wp-content\/uploads\/2023\/09\/image-157-2.png 1285w\" sizes=\"auto, (max-width: 1024px) 100vw, 1024px\" \/><\/a><\/figure>\n<\/div>\n<p class=\"has-base-2-background-color has-background\"><img decoding=\"async\" src=\"https:\/\/s.w.org\/images\/core\/emoji\/14.0.0\/72x72\/1f517.png\" alt=\"\ud83d\udd17\" class=\"wp-smiley\" style=\"height: 1em; max-height: 1em;\" \/> <strong>Recommended<\/strong>: <a href=\"https:\/\/blog.finxter.com\/claude-2-read-ten-papers-in-one-prompt-with-massive-200k-token-context\/?tl_inbound=1&amp;tl_target_all=1&amp;tl_form_type=1&amp;tl_period_type=3\">Claude 2 LLM Reads Ten Papers in One Prompt with Massive 200k Token Context<\/a><\/p>\n<p>The introduction of <a href=\"https:\/\/blog.finxter.com\/ways-to-get-started-with-llama2\/\">LLaMA 1 and 2<\/a> was a significant step in improving the quality of open-source LLMs. LLaMA is a suite of different LLMs with sizes ranging from 7 billion to 65 billion parameters. These models strike a balance between performance and inference efficiency. <\/p>\n<div class=\"wp-block-image\">\n<figure class=\"aligncenter size-full\"><img decoding=\"async\" loading=\"lazy\" width=\"1002\" height=\"926\" src=\"https:\/\/blog.finxter.com\/wp-content\/uploads\/2023\/09\/image-244.png\" alt=\"\" class=\"wp-image-1651784\" srcset=\"https:\/\/blog.finxter.com\/wp-content\/uploads\/2023\/09\/image-244.png 1002w, https:\/\/blog.finxter.com\/wp-content\/uploads\/2023\/09\/image-244-300x277.png 300w, https:\/\/blog.finxter.com\/wp-content\/uploads\/2023\/09\/image-244-768x710.png 768w\" sizes=\"auto, (max-width: 1002px) 100vw, 1002px\" \/><\/figure>\n<\/div>\n<p><a href=\"https:\/\/blog.finxter.com\/llama-2-how-metas-free-open-source-llm-beats-gpt-4\/\">LLaMA models<\/a> are pre-trained on a corpus containing over 1.4 trillion tokens of text, making it one of the largest open-source datasets available. The release of LLaMA models sparked an explosion of open-source research and development in the LLM community.<\/p>\n<p>Here&#8217;s a couple of open-source LLMs that were kicked off after the release of Llama: <a href=\"https:\/\/blog.finxter.com\/gpt4all-vs-alpaca-comparing-open-source-llms\/\">Alpaca<\/a>, <a href=\"https:\/\/blog.finxter.com\/gpt4all-vs-vicuna-battle-of-open-source-llms-%e2%9a%94%ef%b8%8f\/\">Vicuna<\/a>, <a href=\"https:\/\/blog.finxter.com\/choose-the-best-open-source-llm-with-this-powerful-tool\/\">Koala<\/a>, <a href=\"https:\/\/blog.finxter.com\/gpt4all-vs-vicuna-battle-of-open-source-llms-%e2%9a%94%ef%b8%8f\/\">GPT4All<\/a>:<\/p>\n<div class=\"wp-block-image\">\n<figure class=\"aligncenter size-full\"><a href=\"https:\/\/cameronrwolfe.substack.com\/p\/the-history-of-open-source-llms-better\"><img decoding=\"async\" loading=\"lazy\" width=\"873\" height=\"477\" src=\"https:\/\/blog.finxter.com\/wp-content\/uploads\/2023\/09\/image-116.png\" alt=\"\" class=\"wp-image-1651787\" srcset=\"https:\/\/blog.finxter.com\/wp-content\/uploads\/2023\/09\/image-116.png 873w, https:\/\/blog.finxter.com\/wp-content\/uploads\/2023\/09\/image-116-300x164.png 300w, https:\/\/blog.finxter.com\/wp-content\/uploads\/2023\/09\/image-116-768x420.png 768w\" sizes=\"auto, (max-width: 873px) 100vw, 873px\" \/><\/a><figcaption class=\"wp-element-caption\"><a href=\"https:\/\/cameronrwolfe.substack.com\/p\/the-history-of-open-source-llms-better\">Image source<\/a><\/figcaption><\/figure>\n<\/div>\n<p>LLaMA-2, the latest release, sets a new state-of-the-art among open-source LLMs. These models are pre-trained on 2 trillion tokens of publicly available data and utilize a novel approach called Grouped Query Attention (GQA) to improve inference efficiency.<\/p>\n<div class=\"wp-block-image\">\n<figure class=\"aligncenter size-full\"><img decoding=\"async\" loading=\"lazy\" width=\"1024\" height=\"556\" src=\"https:\/\/blog.finxter.com\/wp-content\/uploads\/2023\/09\/image-64-1024x556-1.png\" alt=\"\" class=\"wp-image-1651785\" srcset=\"https:\/\/blog.finxter.com\/wp-content\/uploads\/2023\/09\/image-64-1024x556-1.png 1024w, https:\/\/blog.finxter.com\/wp-content\/uploads\/2023\/09\/image-64-1024x556-1-300x163.png 300w, https:\/\/blog.finxter.com\/wp-content\/uploads\/2023\/09\/image-64-1024x556-1-768x417.png 768w\" sizes=\"auto, (max-width: 1024px) 100vw, 1024px\" \/><\/figure>\n<\/div>\n<p><a href=\"https:\/\/blog.finxter.com\/mpt-7b-llm-quick-guide\/\">MPT<\/a>, another commercially-usable open-source LLM suite, was released by MosaicML. MPT-7B and <a href=\"https:\/\/blog.finxter.com\/six-best-private-secure-llms-in-2023\/\">MPT-30B<\/a> models gained popularity due to their performance and ability to be used in commercial applications. While these models perform slightly worse than proprietary models like GPT-based variants, they outperform other open-source models.<\/p>\n<p><a href=\"https:\/\/blog.finxter.com\/six-best-private-secure-llms-in-2023\/\">Falcon<\/a>, an open-source alternative to proprietary models, was the first to match the quality of closed-source LLMs. Falcon-7B and Falcon-40B models are commercially licensed and perform exceptionally well. They are pre-trained on a custom-curated corpus called RefinedWeb, which contains over 5 trillion tokens of text.<\/p>\n<div class=\"wp-block-image\">\n<figure class=\"aligncenter\"><img decoding=\"async\" loading=\"lazy\" width=\"1024\" height=\"909\" src=\"https:\/\/blog.finxter.com\/wp-content\/uploads\/2023\/09\/image-66-1024x909.png\" alt=\"\" class=\"wp-image-1651491\" srcset=\"https:\/\/blog.finxter.com\/wp-content\/uploads\/2023\/09\/image-66-1024x909.png 1024w, https:\/\/blog.finxter.com\/wp-content\/uploads\/2023\/09\/image-66-300x266.png 300w, https:\/\/blog.finxter.com\/wp-content\/uploads\/2023\/09\/image-66-768x681.png 768w, https:\/\/blog.finxter.com\/wp-content\/uploads\/2023\/09\/image-66.png 1162w\" sizes=\"auto, (max-width: 1024px) 100vw, 1024px\" \/><\/figure>\n<\/div>\n<p>You can currently try the Falcon-180B Demo <a href=\"https:\/\/huggingface.co\/spaces\/tiiuae\/falcon-180b-demo\">here<\/a>.<\/p>\n<p class=\"has-global-color-8-background-color has-background\"><img decoding=\"async\" src=\"https:\/\/s.w.org\/images\/core\/emoji\/14.0.0\/72x72\/1f4c8.png\" alt=\"\ud83d\udcc8\" class=\"wp-smiley\" style=\"height: 1em; max-height: 1em;\" \/> <strong>TLDR<\/strong>: Open-source LLMs include OPT, BLOOM, LLaMa, MPT, and Falcon, each pre-trained on extensive tokens. LLaMa-2 and Falcon stand out for their innovative approaches and extensive training data. <\/p>\n<p><img decoding=\"async\" src=\"https:\/\/s.w.org\/images\/core\/emoji\/14.0.0\/72x72\/1f449.png\" alt=\"\ud83d\udc49\" class=\"wp-smiley\" style=\"height: 1em; max-height: 1em;\" \/> <strong><em>For the best open-source LLM, consider using <a href=\"https:\/\/huggingface.co\/lmsys\/vicuna-33b-v1.3\">Vicuna-33B<\/a> for its superior performance among non-commercial options.<\/em><\/strong><\/p>\n<p>Also, make sure to check out my other article on the Finxter blog: <img decoding=\"async\" src=\"https:\/\/s.w.org\/images\/core\/emoji\/14.0.0\/72x72\/1f447.png\" alt=\"\ud83d\udc47\" class=\"wp-smiley\" style=\"height: 1em; max-height: 1em;\" \/><\/p>\n<div class=\"wp-block-image\">\n<figure class=\"aligncenter size-full\"><a href=\"https:\/\/blog.finxter.com\/six-best-private-secure-llms-in-2023\/\" target=\"_blank\" rel=\"noreferrer noopener\"><img decoding=\"async\" loading=\"lazy\" width=\"1024\" height=\"575\" src=\"https:\/\/blog.finxter.com\/wp-content\/uploads\/2023\/09\/image-67-1024x575-1.png\" alt=\"\" class=\"wp-image-1651786\" srcset=\"https:\/\/blog.finxter.com\/wp-content\/uploads\/2023\/09\/image-67-1024x575-1.png 1024w, https:\/\/blog.finxter.com\/wp-content\/uploads\/2023\/09\/image-67-1024x575-1-300x168.png 300w, https:\/\/blog.finxter.com\/wp-content\/uploads\/2023\/09\/image-67-1024x575-1-768x431.png 768w\" sizes=\"auto, (max-width: 1024px) 100vw, 1024px\" \/><\/a><\/figure>\n<\/div>\n<p class=\"has-base-2-background-color has-background\"><img decoding=\"async\" src=\"https:\/\/s.w.org\/images\/core\/emoji\/14.0.0\/72x72\/1f517.png\" alt=\"\ud83d\udd17\" class=\"wp-smiley\" style=\"height: 1em; max-height: 1em;\" \/> <strong>Recommended<\/strong>: <a href=\"https:\/\/blog.finxter.com\/six-best-private-secure-llms-in-2023\/\">Six Best Private &amp; Secure LLMs in 2023<\/a><\/p>\n<p>The post <a rel=\"nofollow\" href=\"https:\/\/blog.finxter.com\/5-best-open-source-llms-in-2023-two-minute-guide\/\">5 Best Open-Source LLMs in 2023 (Two-Minute Guide)<\/a> appeared first on <a rel=\"nofollow\" href=\"https:\/\/blog.finxter.com\">Be on the Right Side of Change<\/a>.<\/p>\n","protected":false},"excerpt":{"rendered":"<p>5\/5 &#8211; (1 vote) Open-source research on large language models (LLMs) is crucial for democratizing this powerful technology. Although open-source LLMs are now widely used and studied, they faced initial challenges and criticism. Early attempts at creating open-source LLMs like OPT and BLOOM had poor performance compared to closed-source models. This led researchers to realize [&hellip;]<\/p>\n","protected":false},"author":2,"featured_media":0,"comment_status":"open","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[857],"tags":[73,468,528],"class_list":["post-134949","post","type-post","status-publish","format-standard","hentry","category-python-tut","tag-programming","tag-python","tag-tutorial"],"_links":{"self":[{"href":"https:\/\/sickgaming.net\/blog\/wp-json\/wp\/v2\/posts\/134949","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/sickgaming.net\/blog\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/sickgaming.net\/blog\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/sickgaming.net\/blog\/wp-json\/wp\/v2\/users\/2"}],"replies":[{"embeddable":true,"href":"https:\/\/sickgaming.net\/blog\/wp-json\/wp\/v2\/comments?post=134949"}],"version-history":[{"count":0,"href":"https:\/\/sickgaming.net\/blog\/wp-json\/wp\/v2\/posts\/134949\/revisions"}],"wp:attachment":[{"href":"https:\/\/sickgaming.net\/blog\/wp-json\/wp\/v2\/media?parent=134949"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/sickgaming.net\/blog\/wp-json\/wp\/v2\/categories?post=134949"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/sickgaming.net\/blog\/wp-json\/wp\/v2\/tags?post=134949"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}