{"id":119244,"date":"2025-05-29T03:48:07","date_gmt":"2025-05-29T03:48:07","guid":{"rendered":"https:\/\/kamucalisani.net\/?p=119244"},"modified":"2025-05-29T03:48:07","modified_gmt":"2025-05-29T03:48:07","slug":"amazon-cok-modlu-yapay-zeka-ailesi-novayi-duyurdu-openaia-buyuk-rakip","status":"publish","type":"post","link":"https:\/\/kamucalisani.net\/index.php\/2025\/05\/29\/amazon-cok-modlu-yapay-zeka-ailesi-novayi-duyurdu-openaia-buyuk-rakip\/","title":{"rendered":"Amazon, \u00e7ok modlu yapay zeka ailesi Nova\u2019y\u0131 duyurdu: OpenAI’a b\u00fcy\u00fck rakip"},"content":{"rendered":"

<\/span> Teknoloji devi Amazon<\/strong>‘un bulut bili\u015fim b\u00f6l\u00fcm\u00fc olan Amazon Web Services (AWS), d\u00fczenledi\u011fi re:Invent konferans\u0131nda AWS ekosistemine entegre edece\u011fi yeni \u00e7ok modlu<\/strong> \u00fcretken yapay zeka<\/strong> modelleri ailesini duyurdu. Nova <\/strong>ad\u0131n\u0131n verildi\u011fi yapay zeka ailesi Amazon\u2019un yapay zeka alan\u0131ndaki yetkinliklerini g\u00fc\u00e7lendirme ve rekabette \u00f6ne ge\u00e7me stratejisinin bir par\u00e7as\u0131 olarak dikkat \u00e7ekiyor. <\/figure>\n

Her i\u015fe uygun Nova\u2019lar var<\/b><\/p>\n

Modellere ve detaylar\u0131na ge\u00e7meden \u00f6nce hemen bir \u00f6rnek vermek istiyorum. Hemen \u00fcstteki GIF’te Amazon Nova Pro<\/strong> ile olu\u015fturulmu\u015f bir Amerikan futbolu sahnesi bulunuyor. Fizik ve tutarl\u0131l\u0131k konusunda model burada \u00e7ok g\u00fc\u00e7l\u00fc g\u00f6r\u00fcn\u00fcyor. Ba\u015fka bir \u00f6rnek daha var. Bu \u00f6rnekte ise gelece\u011fin reklamlar\u0131 g\u00f6z\u00fcm\u00fcze sokuluyor:\u00a0Amazon Ads<\/strong> taraf\u0131ndan Amazon Nova Reel <\/strong>ile olu\u015fturulan Pasta City.<\/p>\n

Amazon\u2019un blog yaz\u0131s\u0131nda yer alan bilgilere g\u00f6re, Nova serisi \u015fu anda \u00fc\u00e7 farkl\u0131 \u201canlama\u201d modelini i\u00e7eriyor. Nova Micro<\/strong>, h\u0131z ve maliyet optimizasyonuyla \u00f6ne \u00e7\u0131kan bir metin model konumundayken Nova Lite<\/strong>, g\u00f6rsel, video ve metin girdileri alarak metin \u00fcretebilen d\u00fc\u015f\u00fck maliyetli \u00e7ok modlu bir model. Ayr\u0131ca daha karma\u015f\u0131k ve yetenekli olan \u00e7ok modlu Nova<\/strong> Pro<\/strong> modeli de bulunuyor. Bu modeller 128 bin ile 300 bin token\u2019lik (100 bin ila 225 bin kelime) ba\u011flam penceresine sahip. Ancak ba\u011flam pencereleri gelecek y\u0131l 2 milyonu ge\u00e7ecek.<\/p>\n

\u015eirket ayr\u0131ca Nova Premier<\/strong> adl\u0131, ileri d\u00fczeyde karma\u015f\u0131k muhakeme g\u00f6revleri i\u00e7in geli\u015ftirdi\u011fi bir modeli 2025\u2019in ba\u015flar\u0131nda kullan\u0131ma sunmay\u0131 hedefliyor. Bunun yan\u0131 s\u0131ra, Amazon, i\u00e7erik \u00fcretimine y\u00f6nelik iki yeni model daha tan\u0131tt\u0131. G\u00f6rsel i\u00e7erik \u00fcretimi i\u00e7in optimize edilmi\u015f bir model olan Nova Canvas<\/strong> ve viideo \u00fcretiminde kullan\u0131lacak Nova<\/strong> Reel<\/strong>. Bu modeller, \u201csorumlu yapay zeka kullan\u0131m\u0131\u201d te\u015fvik etmek amac\u0131yla dijital filigran ekleme \u00f6zellikleriyle donat\u0131lm\u0131\u015f durumda. Bu modeller kullan\u0131ma sunuldu.<\/p>\n

Amazon, 2025 y\u0131l\u0131 i\u00e7inde konu\u015fmadan konu\u015fmaya \u00e7eviri<\/strong> yapabilen ve tamamen \u00e7ok modlu girdilerle \u00e7al\u0131\u015fabilen yerel bir model de geli\u015ftirece\u011fini a\u00e7\u0131klad\u0131. Amazon, konu\u015fmadan konu\u015fmaya modelinin ayr\u0131ca ton ve tempo gibi s\u00f6zl\u00fc ve s\u00f6zs\u00fcz ipu\u00e7lar\u0131n\u0131 yorumlayabilece\u011fini ve do\u011fal, \u201cinsan benzeri\u201d sesler sunabilece\u011fini s\u00f6yl\u00fcyor. Bu, OpenaI\u2019\u0131n GPT-4o ile sundu\u011fu do\u011fal konu\u015fmaya benzeyecek.<\/p>\n

<\/span> Bir de Any-to-Any (Herhangi Birinden Herhangi Birine diye \u00e7evirsek \u00e7ok yanl\u0131\u015f olmaz) modeli haz\u0131rlan\u0131yor. Bu model, teorik olarak \u00e7evirmenlerden i\u00e7erik edit\u00f6rlerine ve yapay zeka asistanlar\u0131na kadar uygulamalar\u0131 g\u00fc\u00e7lendirecek. Bu modele metin, konu\u015fma, g\u00f6r\u00fcnt\u00fc veya video girebilecek ve metin, konu\u015fma, g\u00f6r\u00fcnt\u00fc veya video \u00e7\u0131kt\u0131s\u0131 al\u0131nabilecek; yani herhangi birinden herhangi birine d\u00f6n\u00fc\u015f\u00fcm yap\u0131lacak. <\/figure>\n

Anthropic yat\u0131r\u0131mlar\u0131 meyve verdi<\/b><\/p>\n

Amazon, yapay zeka altyap\u0131s\u0131n\u0131 geli\u015ftirmek i\u00e7in Anthropic ile 8 milyar dolarl\u0131k bir yat\u0131r\u0131m kapsam\u0131nda ortakl\u0131k kurdu. Bu i\u015f birli\u011fi \u00e7er\u00e7evesinde, Amazon\u2019un Trainium 2 \u00e7iplerine dayal\u0131 d\u00fcnyan\u0131n en b\u00fcy\u00fck yapay zeka hesaplama k\u00fcmesinin in\u015fas\u0131na ba\u015fland\u0131. Bu devasa altyap\u0131, Anthropic’in gelecekteki modellerini olu\u015fturup da\u011f\u0131tmas\u0131na olanak sa\u011flayacak. Amazon ise Anthropic teknolojilerinden yararlanacak. Bu, OpenAI ve Microsoft\u2019un yak\u0131n ili\u015fkisinin bir benzeri.<\/p>\n\n

Kaynak\u00a0 :\u00a0https:\/\/www.donanimhaber.com\/amazon-cok-modlu-yapay-zeka-ailesi-nova-yi-duyurdu–184937<\/span><\/p>\n","protected":false},"excerpt":{"rendered":"

Teknoloji devi Amazon’un bulut bili\u015fim b\u00f6l\u00fcm\u00fc olan Amazon Web Services (AWS), d\u00fczenledi\u011fi re:Invent konferans\u0131nda AWS ekosistemine entegre edece\u011fi yeni \u00e7ok modlu \u00fcretken yapay zeka modelleri ailesini duyurdu. Nova ad\u0131n\u0131n verildi\u011fi yapay zeka …<\/p>\n","protected":false},"author":1,"featured_media":119245,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[8],"tags":[533,4539,449,7924,571],"class_list":["post-119244","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-teknoloji","tag-amazon","tag-metin","tag-model","tag-nova","tag-yapay-zeka"],"_links":{"self":[{"href":"https:\/\/kamucalisani.net\/index.php\/wp-json\/wp\/v2\/posts\/119244","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/kamucalisani.net\/index.php\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/kamucalisani.net\/index.php\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/kamucalisani.net\/index.php\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/kamucalisani.net\/index.php\/wp-json\/wp\/v2\/comments?post=119244"}],"version-history":[{"count":1,"href":"https:\/\/kamucalisani.net\/index.php\/wp-json\/wp\/v2\/posts\/119244\/revisions"}],"predecessor-version":[{"id":119248,"href":"https:\/\/kamucalisani.net\/index.php\/wp-json\/wp\/v2\/posts\/119244\/revisions\/119248"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/kamucalisani.net\/index.php\/wp-json\/wp\/v2\/media\/119245"}],"wp:attachment":[{"href":"https:\/\/kamucalisani.net\/index.php\/wp-json\/wp\/v2\/media?parent=119244"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/kamucalisani.net\/index.php\/wp-json\/wp\/v2\/categories?post=119244"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/kamucalisani.net\/index.php\/wp-json\/wp\/v2\/tags?post=119244"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}