{"id":1902,"date":"2023-12-13T09:53:17","date_gmt":"2023-12-13T01:53:17","guid":{"rendered":"https:\/\/www.1ai.net\/?p=1902"},"modified":"2023-12-13T09:53:17","modified_gmt":"2023-12-13T01:53:17","slug":"%e5%be%ae%e8%bd%af%e5%8f%91%e5%b8%83%e5%b0%8f%e5%9e%8b%e8%af%ad%e8%a8%80-ai-%e6%a8%a1%e5%9e%8b-phi-2%ef%bc%8c%e6%80%a7%e8%83%bd%e4%bc%98%e4%ba%8e-llama-2%e3%80%81mistral-7b","status":"publish","type":"post","link":"https:\/\/www.1ai.net\/en\/1902.html","title":{"rendered":"Microsoft releases small language AI model Phi-2, which outperforms Llama 2 and Mistral 7B"},"content":{"rendered":"<p><a href=\"https:\/\/www.1ai.net\/en\/tag\/%e5%be%ae%e8%bd%af%e7%a0%94%e7%a9%b6%e9%99%a2\" title=\"[Sees articles with labels]\" target=\"_blank\" >Microsoft Research<\/a>Local time Tuesday announced its<strong> <a href=\"https:\/\/www.1ai.net\/en\/tag\/phi-2\" title=\"_Other Organiser\" target=\"_blank\" >Phi-2<\/a> <a href=\"https:\/\/www.1ai.net\/en\/tag\/%e5%b0%8f%e5%9e%8b%e8%af%ad%e8%a8%80%e6%a8%a1%e5%9e%8b\" title=\"[Sees articles with [small language model] labels]\" target=\"_blank\" >Small Language Model<\/a>(SML)<\/strong>This is<strong>A text-to-text artificial intelligence<\/strong>program, which, according to a post on the X platform, it<strong>\"Small enough to run on a laptop or mobile device.\"<\/strong>.<\/p>\n<p class=\"article-content__img\"><img loading=\"lazy\" decoding=\"async\" class=\"alignnone size-full wp-image-1903\" src=\"https:\/\/www.1ai.net\/wp-content\/uploads\/2023\/12\/6383805386077404889902939.jpg\" alt=\"\" width=\"1000\" height=\"1134\" \/><\/p>\n<p>With 2.7 billion parameters, Phi-2 rivals the performance of other larger models, including Meta's\u00a0<strong>Llama 2-7B<\/strong>(with 7 billion parameters) and Mistral-7B (another 7 billion parameter model).<\/p>\n<p>Microsoft Research has also released in its Phi-2<strong>Blog Posts<\/strong>It is noted that although Phi-2 has fewer parameters than Google's brand new\u00a0<strong>Gemini<\/strong>\u00a0Nano 2 model (which has more than 5 billion parameters), but it performs better and has less 'toxicity' and bias in its responses than Llama 2.<\/p>\n<p>Microsoft also took the opportunity to take a little dig at Google's release of the Gemini model, which is on the scale of its upcoming<span class=\"spamTxt\">maximum<\/span>Competencies<span class=\"spamTxt\">Strongest<\/span>'s new AI model, Gemini Ultra, in a demo video showing its ability to solve relatively complex physics problems and even correct student errors. It turns out that while Phi-2 may be a fraction of the size of Gemini Ultra, it can also answer questions correctly and correct students using the same prompts.<\/p>\n<p>However, as encouraging as these findings are, Phi-2 currently has one major limitation: according to the Microsoft Research license, the<strong>It is for \"research purposes only\" and not for commercial use.<\/strong>The license further states that Phi-2 may only be used for \"non-commercial, non-revenue producing research purposes. The license further specifies that Phi-2 can only be used for \"non-commercial, non-revenue generating research purposes.\" Therefore, companies wishing to build products based on it are out of luck for the time being.<\/p>","protected":false},"excerpt":{"rendered":"<p>Microsoft Research announced Tuesday local time its Phi-2 Small Language Model (SML), a text-to-text AI program that is \"small enough to run on a laptop or mobile device,\" according to a post on the X platform. With 2.7 billion parameters, Phi-2 rivals the performance of other, larger models, including Meta's Llama 2-7B (with 7 billion parameters) and Mistral-7B (another 7 billion parameter model). Microsoft Research also noted in its Phi-2 release blog post that although Phi-2 has fewer parameters than Google's new Gemini Nano 2 model (which has more than 5 billion parameters), it outperforms it by a mile!<\/p>","protected":false},"author":1,"featured_media":0,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[146],"tags":[632,634,633],"collection":[],"class_list":["post-1902","post","type-post","status-publish","format-standard","hentry","category-news","tag-phi-2","tag-634","tag-633"],"acf":[],"_links":{"self":[{"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/posts\/1902","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/comments?post=1902"}],"version-history":[{"count":0,"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/posts\/1902\/revisions"}],"wp:attachment":[{"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/media?parent=1902"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/categories?post=1902"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/tags?post=1902"},{"taxonomy":"collection","embeddable":true,"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/collection?post=1902"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}