{"id":3209890,"date":"2025-01-29T12:53:02","date_gmt":"2025-01-29T11:53:02","guid":{"rendered":"https:\/\/kallxo.com\/?p=3209890"},"modified":"2025-01-29T12:53:02","modified_gmt":"2025-01-29T11:53:02","slug":"alibaba-nxjerr-modelin-e-vet-te-ai-se-thote-se-ia-kalon-deepseekut","status":"publish","type":"post","link":"https:\/\/www.drakonispahiu.com\/?p=3209890","title":{"rendered":"Alibaba nxjerr modelin e vet t\u00eb AI-s\u00eb – thot\u00eb se ia kalon DeepSeekut"},"content":{"rendered":"

Kompania kineze e teknologjis\u00eb “Alibaba” ka nxjerr\u00eb t\u00eb m\u00ebrkur\u00ebn, m\u00eb 29.01.2025, nj\u00eb version t\u00eb ri t\u00eb modelit t\u00eb saj t\u00eb Inteligjenc\u00ebs Artificiale Qwen 2.5 p\u00ebr t\u00eb cilin \u00a0se tejkalonte DeepSeek-V3, tashm\u00eb shum\u00eb t\u00eb vler\u00ebsuar – raporton Agjencia amerikane e lajmeve Reuters.<\/a><\/p>\n

Koha e pazakonshme e nxjerrjes n\u00eb treg t\u00eb Qwen 2.5-Max, n\u00eb dit\u00ebn e par\u00eb t\u00eb Vitit t\u00eb Ri H\u00ebnor, kur shumica e kinez\u00ebve jan\u00eb n\u00eb pushime dhe me familjet e tyre, tregon presionin q\u00eb rritja meteorike e Startupit kinez t\u00eb Inteligjenc\u00ebs Artificiale, DeepSeek n\u00eb tre jav\u00ebt e fundit \u00ebsht\u00eb ushtruar jo vet\u00ebm ndaj rival\u00ebve jasht\u00eb shtetit, por edhe ndaj konkurrenc\u00ebs vendase.<\/p>\n

“Qwen 2.5-Max performon m\u00eb mir\u00eb pothuajse n\u00eb t\u00eb gjith\u00eb bordin GPT-4o, DeepSeek-V3 dhe Llama-3.1-405B.\u201d – thuhet n\u00eb njoftimin e Nj\u00ebsis\u00eb Cloud t\u00eb Alibabas, postuar n\u00eb llogarin\u00eb e zyrtare WeChat – duke iu referuar OpenAI dhe modeleve m\u00eb t\u00eb avancuara t\u00eb Inteligjenc\u00ebs Artificiale t\u00eb Meta-s.<\/p>\n

Publikimi i 10 janarit i Inteligjenc\u00ebs Artificiale t\u00eb DeepSeek, mund\u00ebsuar nga modeli DeepSeek-V3, si dhe publikimi i modelit t\u00eb tij R1 m\u00eb 20 janar, ka tronditur Lugin\u00ebn e Silikonit (n\u00eb Shtetet e Bashkuara) dhe ka shkaktuar r\u00ebnien e aksioneve t\u00eb teknologjis\u00eb, me zhvillimin e supozuar t\u00eb lir\u00eb t\u00eb Startupit kinez dhe kostot e p\u00ebrdorimit q\u00eb i kan\u00eb b\u00ebr\u00eb investitor\u00ebt t\u00eb v\u00ebn\u00eb n\u00eb dyshim planet e m\u00ebdha t\u00eb shpenzimeve nga firmat kryesore t\u00eb AI n\u00eb Shtetet e Bashkuara.<\/p>\n

Por, suksesi i DeepSeek ka nxitur gjithashtu p\u00ebrplasje midis konkurrent\u00ebve t\u00eb tij vendas p\u00ebr t\u00eb p\u00ebrmir\u00ebsuar modelet e tyre t\u00eb Inteligjenc\u00ebs Artificiale.<\/p>\n

Dy dit\u00eb pas l\u00ebshimit t\u00eb DeepSeek-R1, pronari i TikTok ByteDance publikoi \u200b\u200b\u200b\u200bnj\u00eb p\u00ebrdit\u00ebsim t\u00eb modelit t\u00eb tij t\u00eb Inteligjenc\u00ebs Artificiale, p\u00ebr t\u00eb cilin pretendoi se ishte m\u00eb i mir\u00eb se O1 i OpenAI, mb\u00ebshtetur nga Microsoft n\u00eb AIME, nj\u00eb test standard q\u00eb mat se sa mir\u00eb modelet e AI kuptojn\u00eb dhe u p\u00ebrgjigjen udh\u00ebzimeve komplekse.<\/p>\n

Kjo i dha jehon\u00eb pretendimit t\u00eb DeepSeek se modeli i tij R1 rivalizonte o1 t\u00eb OpenAI, n\u00eb disa standarde t\u00eb performanc\u00ebs.<\/p>\n

 <\/p>\n

DeepSeek karshi konkurrent\u00ebve vendas<\/strong><\/p>\n

Paraardh\u00ebsi i modelit V3 t\u00eb DeepSeek, DeepSeek-V2 nxiti luft\u00eb \u00e7mimesh t\u00eb modelit t\u00eb AI n\u00eb Kin\u00eb pas publikimit majin e kaluar.<\/p>\n

Fakti q\u00eb DeepSeek-V2 ishte me burim t\u00eb hapur dhe jasht\u00ebzakonisht t\u00eb lir\u00eb, vet\u00ebm 1 juan (0,14 dollar\u00eb) p\u00ebr 1 milion argumente – ose Nj\u00ebsi t\u00eb t\u00eb dh\u00ebnash t\u00eb p\u00ebrpunuara nga modeli i AI – b\u00ebri q\u00eb Nj\u00ebsia Cloud e Alibabas t\u00eb shpallte ulje \u00e7mimi deri n\u00eb 97% n\u00eb gam\u00ebn e modeleve.<\/p>\n

Kompanit\u00eb e tjera kineze t\u00eb teknologjis\u00eb ndoq\u00ebn t\u00eb nj\u00ebjtin shembull, duke p\u00ebrfshir\u00eb Baidu (9888.HK), me nxjerrjen e \u200b\u200bekuivalentit t\u00eb par\u00eb t\u00eb Kin\u00ebs me ChatGPT n\u00eb mars 2023 dhe Kompanis\u00eb m\u00eb t\u00eb vlefshme t\u00eb internetit n\u00eb vend Tencent.<\/p>\n

Liang Wenfeng, themeluesi enigmatik i DeepSeek, ka deklaruar n\u00eb nj\u00eb intervist\u00eb t\u00eb rrall\u00eb p\u00ebr mediumin kinez Waves n\u00eb korrik se Startupit “nuk i interesonte” lufta e \u00e7mimeve dhe se arritja e AGI (Inteligjenc\u00ebs s\u00eb p\u00ebrgjithshme Artificiale) ishte q\u00ebllimi i saj kryesor.<\/p>\n

OpenAI e p\u00ebrcakton AGI-n\u00eb si sisteme autonome, q\u00eb i tejkalojn\u00eb njer\u00ebzit n\u00eb detyrat m\u00eb t\u00eb vlefshme ekonomikisht.<\/p>\n

Nd\u00ebrsa Kompanit\u00eb e m\u00ebdha kineze t\u00eb teknologjis\u00eb si “Alibaba” kan\u00eb qindra-mij\u00ebra punonj\u00ebs. DeepSeek funksionon si nj\u00eb Laborator k\u00ebrkimor, i pajisur kryesisht nga t\u00eb rinj t\u00eb diplomuar dhe student\u00eb t\u00eb doktorat\u00ebs nga Universitetet m\u00eb t\u00eb mira kineze.<\/p>\n

Liang ka theksuar n\u00eb intervist\u00ebn e tij t\u00eb korrikut se beson se Kompanit\u00eb m\u00eb t\u00eb m\u00ebdha t\u00eb teknologjis\u00eb n\u00eb Kin\u00eb mund t\u00eb mos ishin t\u00eb p\u00ebrshtatshme p\u00ebr t\u00eb ardhmen e industris\u00eb s\u00eb AI-s\u00eb – duke krahasuar kostot e tyre t\u00eb larta dhe strukturat e komandimit nga lart-posht\u00eb me funksionimin e dob\u00ebt t\u00eb DeepSeek dhe stilin m\u00eb t\u00eb but\u00eb t\u00eb menaxhimit.<\/p>\n

“Modelet e m\u00ebdha themelore k\u00ebrkojn\u00eb inovacion t\u00eb vazhduesh\u00ebm. Aft\u00ebsit\u00eb e gjigant\u00ebve t\u00eb teknologjis\u00eb kan\u00eb limitet e tyre.” – ka deklaruar ai.<\/p>\n

 <\/p>\n

P\u00ebrgatiti: Rexhep Maloku – Kallxo.com<\/strong><\/p>\n","protected":false},"excerpt":{"rendered":"

Kompania kineze e teknologjis\u00eb “Alibaba” ka nxjerr\u00eb t\u00eb m\u00ebrkur\u00ebn, m\u00eb 29.01.2025, nj\u00eb version t\u00eb ri t\u00eb modelit t\u00eb saj t\u00eb Inteligjenc\u00ebs Artificiale Qwen 2.5 p\u00ebr t\u00eb cilin \u00a0se tejkalonte DeepSeek-V3, tashm\u00eb shum\u00eb t\u00eb vler\u00ebsuar – raporton Agjencia amerikane e lajmeve Reuters. Koha e pazakonshme e nxjerrjes n\u00eb treg t\u00eb Qwen 2.5-Max, n\u00eb dit\u00ebn e par\u00eb […]<\/p>\n","protected":false},"author":9,"featured_media":2494199,"comment_status":"closed","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[29,11],"tags":[2658,2659],"class_list":["post-3209890","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-bote","category-lajm","tag-gara","tag-inteligjenca-artificiale"],"lang":"sq","translations":{"sq":3209890},"pll_sync_post":[],"_links":{"self":[{"href":"https:\/\/www.drakonispahiu.com\/index.php?rest_route=\/wp\/v2\/posts\/3209890","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.drakonispahiu.com\/index.php?rest_route=\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.drakonispahiu.com\/index.php?rest_route=\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.drakonispahiu.com\/index.php?rest_route=\/wp\/v2\/users\/9"}],"replies":[{"embeddable":true,"href":"https:\/\/www.drakonispahiu.com\/index.php?rest_route=%2Fwp%2Fv2%2Fcomments&post=3209890"}],"version-history":[{"count":0,"href":"https:\/\/www.drakonispahiu.com\/index.php?rest_route=\/wp\/v2\/posts\/3209890\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/www.drakonispahiu.com\/index.php?rest_route=\/"}],"wp:attachment":[{"href":"https:\/\/www.drakonispahiu.com\/index.php?rest_route=%2Fwp%2Fv2%2Fmedia&parent=3209890"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.drakonispahiu.com\/index.php?rest_route=%2Fwp%2Fv2%2Fcategories&post=3209890"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.drakonispahiu.com\/index.php?rest_route=%2Fwp%2Fv2%2Ftags&post=3209890"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}