{"id":4150,"date":"2026-09-29T18:44:13","date_gmt":"2026-09-29T18:44:13","guid":{"rendered":"https:\/\/netguide.io\/news\/2026\/09\/29\/mit-batte-il-record-di-benchmark-agi-pub-del-20-con-il-nuovo-llm\/"},"modified":"2026-09-29T18:44:13","modified_gmt":"2026-09-29T18:44:13","slug":"mit-batte-il-record-di-benchmark-agi-pub-del-20-con-il-nuovo-llm","status":"publish","type":"post","link":"https:\/\/netguide.io\/news\/it\/2026\/09\/29\/mit-batte-il-record-di-benchmark-agi-pub-del-20-con-il-nuovo-llm\/","title":{"rendered":"MIT batte il record di benchmark AGI-PUB del 20% con il nuovo LLM"},"content":{"rendered":"<div id=\"netgu-2758839940\" class=\"netgu-before-content netgu-entity-placement\"><script async src=\"\/\/pagead2.googlesyndication.com\/pagead\/js\/adsbygoogle.js?client=ca-pub-6258556257245998\" crossorigin=\"anonymous\"><\/script><ins class=\"adsbygoogle\" style=\"display:block;\" data-ad-client=\"ca-pub-6258556257245998\" \ndata-ad-slot=\"3494115342\" \ndata-ad-format=\"auto\"><\/ins>\n<script> \n(adsbygoogle = window.adsbygoogle || []).push({}); \n<\/script>\n<\/div><p class=\"wp-block-paragraph\">Un team di MIT ha fatto progressi significativi nella soluzione astratta delle attivit\u00e0 utilizzando un modello linguistico di 8 miliardi (LLM) e una tecnica innovativa chiamata Test-Time Training (TTT). Con una performance del 61,9% sul benchmark ARC-AGI-PUB, hanno notevolmente superato il record precedente del 42%. Questi risultati sono notevoli in quanto ARC-AGI-PUB \u00e8 una raccolta impegnativa di compiti che richiedono il riconoscimento del pattern visivo e il ragionamento complesso.<\/p>\n<p class=\"wp-block-paragraph\"><strong>Cos&#8217;\u00e8 la formazione del test-tempo (TTT)?<\/strong> TTT \u00e8 una tecnica che permette al modello di effettuare piccole regolazioni durante il tempo di prova per reagire a nuovi compiti. TTT utilizza il contesto dei dati di input per ottimizzare temporaneamente il modello per l&#8217;attivit\u00e0 corrente. Questo approccio differisce dai metodi tradizionali in cui il modello \u00e8 formato per un compito in anticipo. Soprattutto in compiti complessi e nuovi, TTT mostra forti miglioramenti delle prestazioni.<\/p><div id=\"netgu-780710517\" class=\"netgu-content netgu-entity-placement\"><aside class=\"deals-top deals-top--compact\">\n\n\t\t\t<h3 class=\"deals-top__title\">Top-Deals<\/h3>\n\t\n\t\t\t<ul class=\"deals-top__list\">\n\t\t\t\t\t\t\t<li class=\"deals-top__item\">\n\t\t\t\t\t<a class=\"deals-top__link\" href=\"https:\/\/netguide.io\/deals\/de\/deals\/moon-knight-digital-watch-face-fuer-wear-os-4\/\">\n\t\t\t\t\t\t\t\t\t\t\t\t\t<img class=\"deals-top__image\" src=\"https:\/\/netguide.io\/news\/wp-content\/uploads\/sites\/15\/2026\/09\/deal-27-150x150.png\" alt=\"\" width=\"52\" height=\"52\" loading=\"lazy\" \/>\n\t\t\t\t\t\t\n\t\t\t\t\t\t<span class=\"deals-top__body\">\n\t\t\t\t\t\t\t<span class=\"deals-top__name\">MOON KNIGHT Digital Watch Face f\u00fcr Wear OS 4+<\/span>\n\n\t\t\t\t\t\t\t<span class=\"deals-top__meta\">\n\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t<span class=\"deals-top__price\">Gratis<\/span>\n\t\t\t\t\t\t\t\t\n\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t<s>0.89 \u20ac<\/s>\n\t\t\t\t\t\t\t\t\n\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t<span class=\"deals-discount\">-100%<\/span>\n\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t<\/span>\n\t\t\t\t\t\t<\/span>\n\n\t\t\t\t\t\t<span class=\"deals-top__temperature\">\n\t\t\t\t\t\t\t55\u00b0\n\t\t\t\t\t\t<\/span>\n\t\t\t\t\t<\/a>\n\t\t\t\t<\/li>\n\t\t\t\t\t\t\t<li class=\"deals-top__item\">\n\t\t\t\t\t<a class=\"deals-top__link\" href=\"https:\/\/netguide.io\/deals\/de\/deals\/train-sim-world-7-starter-pack-kostenlos-fuer-ps4-ps5-xbox-pc\/\">\n\t\t\t\t\t\t\t\t\t\t\t\t\t<img class=\"deals-top__image\" src=\"https:\/\/netguide.io\/news\/wp-content\/uploads\/sites\/15\/2026\/09\/capsule_616x353-11-150x150.jpg\" alt=\"\" width=\"52\" height=\"52\" loading=\"lazy\" \/>\n\t\t\t\t\t\t\n\t\t\t\t\t\t<span class=\"deals-top__body\">\n\t\t\t\t\t\t\t<span class=\"deals-top__name\">Train Sim World 7: Starter Pack kostenlos f\u00fcr PS4, PS5, Xbox, PC<\/span>\n\n\t\t\t\t\t\t\t<span class=\"deals-top__meta\">\n\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t<span class=\"deals-top__price\">Gratis<\/span>\n\t\t\t\t\t\t\t\t\n\t\t\t\t\t\t\t\t\n\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t<\/span>\n\t\t\t\t\t\t<\/span>\n\n\t\t\t\t\t\t<span class=\"deals-top__temperature\">\n\t\t\t\t\t\t\t55\u00b0\n\t\t\t\t\t\t<\/span>\n\t\t\t\t\t<\/a>\n\t\t\t\t<\/li>\n\t\t\t\t\t\t\t<li class=\"deals-top__item\">\n\t\t\t\t\t<a class=\"deals-top__link\" href=\"https:\/\/netguide.io\/deals\/de\/deals\/ml2u-watchfaces-fuer-wearos-im-google-play-store-2\/\">\n\t\t\t\t\t\t\t\t\t\t\t\t\t<img class=\"deals-top__image\" src=\"https:\/\/netguide.io\/news\/wp-content\/uploads\/sites\/15\/2026\/09\/2844734_1-150x150.webp\" alt=\"\" width=\"52\" height=\"52\" loading=\"lazy\" \/>\n\t\t\t\t\t\t\n\t\t\t\t\t\t<span class=\"deals-top__body\">\n\t\t\t\t\t\t\t<span class=\"deals-top__name\">ML2U Watchfaces f\u00fcr WearOS im Google Play Store<\/span>\n\n\t\t\t\t\t\t\t<span class=\"deals-top__meta\">\n\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t<span class=\"deals-top__price\">Gratis<\/span>\n\t\t\t\t\t\t\t\t\n\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t<s>1.39 \u20ac<\/s>\n\t\t\t\t\t\t\t\t\n\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t<span class=\"deals-discount\">-100%<\/span>\n\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t<\/span>\n\t\t\t\t\t\t<\/span>\n\n\t\t\t\t\t\t<span class=\"deals-top__temperature\">\n\t\t\t\t\t\t\t54\u00b0\n\t\t\t\t\t\t<\/span>\n\t\t\t\t\t<\/a>\n\t\t\t\t<\/li>\n\t\t\t\t\t\t\t<li class=\"deals-top__item\">\n\t\t\t\t\t<a class=\"deals-top__link\" href=\"https:\/\/netguide.io\/deals\/de\/deals\/kostenloses-dlc-bundle-fuer-gears-of-war-e-day-liquid-death-bundle\/\">\n\t\t\t\t\t\t\t\t\t\t\t\t\t<img class=\"deals-top__image\" src=\"https:\/\/netguide.io\/news\/wp-content\/uploads\/sites\/15\/2026\/09\/2844571_1-150x150.webp\" alt=\"\" width=\"52\" height=\"52\" loading=\"lazy\" \/>\n\t\t\t\t\t\t\n\t\t\t\t\t\t<span class=\"deals-top__body\">\n\t\t\t\t\t\t\t<span class=\"deals-top__name\">Kostenloses DLC-Bundle f\u00fcr Gears of War E-Day \u2013 Liquid Death Bundle<\/span>\n\n\t\t\t\t\t\t\t<span class=\"deals-top__meta\">\n\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t<span class=\"deals-top__price\">Gratis<\/span>\n\t\t\t\t\t\t\t\t\n\t\t\t\t\t\t\t\t\n\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t<\/span>\n\t\t\t\t\t\t<\/span>\n\n\t\t\t\t\t\t<span class=\"deals-top__temperature\">\n\t\t\t\t\t\t\t54\u00b0\n\t\t\t\t\t\t<\/span>\n\t\t\t\t\t<\/a>\n\t\t\t\t<\/li>\n\t\t\t\t\t\t\t<li class=\"deals-top__item\">\n\t\t\t\t\t<a class=\"deals-top__link\" href=\"https:\/\/netguide.io\/deals\/de\/deals\/kostenlose-emote-im-clash-royale-store\/\">\n\t\t\t\t\t\t\t\t\t\t\t\t\t<img class=\"deals-top__image\" src=\"https:\/\/netguide.io\/news\/wp-content\/uploads\/sites\/15\/2026\/09\/share-image-clashroyale-150x150.jpg\" alt=\"\" width=\"52\" height=\"52\" loading=\"lazy\" \/>\n\t\t\t\t\t\t\n\t\t\t\t\t\t<span class=\"deals-top__body\">\n\t\t\t\t\t\t\t<span class=\"deals-top__name\">Kostenlose Emote im Clash Royale-Store<\/span>\n\n\t\t\t\t\t\t\t<span class=\"deals-top__meta\">\n\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t<span class=\"deals-top__price\">Gratis<\/span>\n\t\t\t\t\t\t\t\t\n\t\t\t\t\t\t\t\t\n\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t<\/span>\n\t\t\t\t\t\t<\/span>\n\n\t\t\t\t\t\t<span class=\"deals-top__temperature\">\n\t\t\t\t\t\t\t54\u00b0\n\t\t\t\t\t\t<\/span>\n\t\t\t\t\t<\/a>\n\t\t\t\t<\/li>\n\t\t\t\t\t<\/ul>\n\n\t\t\t\t\t<p class=\"deals-top__more\">\n\t\t\t\t<a href=\"https:\/\/netguide.io\/deals\/\">Alle Deals ansehen \u2192<\/a>\n\t\t\t<\/p>\n\t\t\t<\/aside>\n<\/div>\n<p class=\"wp-block-paragraph\"><strong>Come ha funzionato la svolta al MIT?<\/strong> Il team ha scoperto che una combinazione di fine-tuning iniziale e l&#8217;applicazione di trasformazioni speciali per testare i dati \u00e8 stata cruciale per il successo. Queste modifiche hanno migliorato la capacit\u00e0 del modello di riconoscere i modelli. Inoltre, \u00e8 stato creato un insieme con approcci di programmazione moderni per aumentare il valore delle prestazioni al 61,9% \u2013 una performance equivalente a quella di un umano medio.<\/p>\n<p class=\"wp-block-paragraph\"><strong>Il significato di questo sviluppo<\/strong> I risultati mostrano che i modelli linguistici possono risolvere compiti astratti anche senza metodi simbolici espliciti.<\/p>\n<p class=\"wp-block-paragraph\">Leggi tutto in <a href=\"https:\/\/ekinakyurek.github.io\/papers\/ttt.pdf\" rel=\"noreferrer noopener\" target=\"_blank\">carta ufficiale<\/a>.<\/p>\n<div id=\"netgu-2786785391\" class=\"netgu-after-content netgu-entity-placement\"><script async src=\"\/\/pagead2.googlesyndication.com\/pagead\/js\/adsbygoogle.js?client=ca-pub-6258556257245998\" crossorigin=\"anonymous\"><\/script><ins class=\"adsbygoogle\" style=\"display:block;\" data-ad-client=\"ca-pub-6258556257245998\" \ndata-ad-slot=\"4559785002\" \ndata-ad-format=\"auto\"><\/ins>\n<script> \n(adsbygoogle = window.adsbygoogle || []).push({}); \n<\/script>\n<\/div>","protected":false},"excerpt":{"rendered":"<p>Un team MIT ha colpito il 61,9% sul benchmark ARC-AGI-PUB con un LLM di 8B-parametro e Test-Time Training, rompendo il record precedente del 42% di quasi 20 punti.<\/p>\n","protected":false},"author":1,"featured_media":1624,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"site-sidebar-layout":"default","site-content-layout":"","ast-site-content-layout":"default","site-content-style":"default","site-sidebar-style":"default","ast-global-header-display":"","ast-banner-title-visibility":"","ast-main-header-display":"","ast-hfb-above-header-display":"","ast-hfb-below-header-display":"","ast-hfb-mobile-header-display":"","site-post-title":"","ast-breadcrumbs-content":"","ast-featured-img":"","footer-sml-layout":"","ast-disable-related-posts":"","theme-transparent-header-meta":"","adv-header-id-meta":"","stick-header-meta":"","header-above-stick-meta":"","header-main-stick-meta":"","header-below-stick-meta":"","astra-migrate-meta-layouts":"default","ast-page-background-enabled":"default","ast-page-background-meta":{"desktop":{"background-color":"var(--ast-global-color-5)","background-image":"","background-repeat":"repeat","background-position":"center center","background-size":"auto","background-attachment":"scroll","background-type":"","background-media":"","overlay-type":"","overlay-color":"","overlay-opacity":"","overlay-gradient":""},"tablet":{"background-color":"","background-image":"","background-repeat":"repeat","background-position":"center center","background-size":"auto","background-attachment":"scroll","background-type":"","background-media":"","overlay-type":"","overlay-color":"","overlay-opacity":"","overlay-gradient":""},"mobile":{"background-color":"","background-image":"","background-repeat":"repeat","background-position":"center center","background-size":"auto","background-attachment":"scroll","background-type":"","background-media":"","overlay-type":"","overlay-color":"","overlay-opacity":"","overlay-gradient":""}},"ast-content-background-meta":{"desktop":{"background-color":"var(--ast-global-color-4)","background-image":"","background-repeat":"repeat","background-position":"center center","background-size":"auto","background-attachment":"scroll","background-type":"","background-media":"","overlay-type":"","overlay-color":"","overlay-opacity":"","overlay-gradient":""},"tablet":{"background-color":"var(--ast-global-color-4)","background-image":"","background-repeat":"repeat","background-position":"center center","background-size":"auto","background-attachment":"scroll","background-type":"","background-media":"","overlay-type":"","overlay-color":"","overlay-opacity":"","overlay-gradient":""},"mobile":{"background-color":"var(--ast-global-color-4)","background-image":"","background-repeat":"repeat","background-position":"center center","background-size":"auto","background-attachment":"scroll","background-type":"","background-media":"","overlay-type":"","overlay-color":"","overlay-opacity":"","overlay-gradient":""}},"footnotes":""},"categories":[],"tags":[],"class_list":["post-4150","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry"],"brizy_media":[],"_links":{"self":[{"href":"https:\/\/netguide.io\/news\/wp-json\/wp\/v2\/posts\/4150","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/netguide.io\/news\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/netguide.io\/news\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/netguide.io\/news\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/netguide.io\/news\/wp-json\/wp\/v2\/comments?post=4150"}],"version-history":[{"count":1,"href":"https:\/\/netguide.io\/news\/wp-json\/wp\/v2\/posts\/4150\/revisions"}],"predecessor-version":[{"id":4151,"href":"https:\/\/netguide.io\/news\/wp-json\/wp\/v2\/posts\/4150\/revisions\/4151"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/netguide.io\/news\/wp-json\/wp\/v2\/media\/1624"}],"wp:attachment":[{"href":"https:\/\/netguide.io\/news\/wp-json\/wp\/v2\/media?parent=4150"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/netguide.io\/news\/wp-json\/wp\/v2\/categories?post=4150"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/netguide.io\/news\/wp-json\/wp\/v2\/tags?post=4150"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}