{"id":1278,"date":"2023-06-05T17:43:13","date_gmt":"2023-06-05T17:43:13","guid":{"rendered":"https:\/\/dailyai.com\/?p=1278"},"modified":"2023-06-05T17:43:13","modified_gmt":"2023-06-05T17:43:13","slug":"new-mit-self-learning-language-models-can-beat-other-llms","status":"publish","type":"post","link":"https:\/\/dailyai.com\/fr\/2023\/06\/new-mit-self-learning-language-models-can-beat-other-llms\/","title":{"rendered":"Les nouveaux mod\u00e8les linguistiques \u00e0 auto-apprentissage du MIT peuvent battre les autres mod\u00e8les linguistiques \u00e0 auto-apprentissage"},"content":{"rendered":"

Au cours des six derniers mois, nous avons assist\u00e9 \u00e0 un boom r\u00e9volutionnaire des <\/span>aliment\u00e9 par l'IA<\/span><\/a> Les LLM (Large Language Models) occupent le devant de la sc\u00e8ne. Mais est-il toujours n\u00e9cessaire qu'un produit ou un service d'IA soit bas\u00e9 sur des LLM ? Selon un article, les nouveaux mod\u00e8les linguistiques auto-apprenants du MIT ne sont pas bas\u00e9s sur des LLM et peuvent surpasser certains des autres grands syst\u00e8mes d'IA qui dominent actuellement le secteur.<\/span><\/p>\n

Un groupe de chercheurs du Computer Science and Artificial Intelligence Laboratory (CSAIL) du MIT a mis au point une nouvelle fa\u00e7on d'aborder les mod\u00e8les linguistiques de l'IA.<\/span><\/p>\n

Il s'agit d'une r\u00e9alisation r\u00e9volutionnaire qui met l'accent sur les mod\u00e8les de langage plus petits et sur leur capacit\u00e9 \u00e0 r\u00e9soudre les probl\u00e8mes d'inefficacit\u00e9, ainsi que les probl\u00e8mes de protection de la vie priv\u00e9e li\u00e9s au d\u00e9veloppement de grands mod\u00e8les d'intelligence artificielle bas\u00e9s sur des donn\u00e9es textuelles.<\/span><\/p>\n

Avec l'\u00e9mergence des <\/span>L'\u00e9quipe d'OpenAI<\/span><\/a> ChatGPT bas\u00e9 sur les mod\u00e8les de langage GPT-3 et GPT-4, de nombreuses entreprises se sont lanc\u00e9es dans la course \u00e0 l'IA, notamment Google Bard, et d'autres entreprises. <\/span>IA g\u00e9n\u00e9rative<\/span><\/a> qui permettent de g\u00e9n\u00e9rer du texte, des images et m\u00eame des vid\u00e9os.<\/span><\/p>\n

Toutefois, pour produire des r\u00e9sultats d'une qualit\u00e9 irr\u00e9prochable, ces syst\u00e8mes s'appuient sur un grand nombre de donn\u00e9es dont le traitement informatique est co\u00fbteux. Nombre de ces syst\u00e8mes importent des donn\u00e9es pour la formation via des API, ce qui comporte des risques tels que les fuites de donn\u00e9es et d'autres probl\u00e8mes li\u00e9s \u00e0 la protection de la vie priv\u00e9e.<\/span><\/p>\n

Encha\u00eenement textuel<\/b><\/h2>\n

Selon un nouveau document intitul\u00e9<\/span> L'entartage en tant qu'auto-apprentissage robuste<\/span><\/a> Dans une \u00e9tude publi\u00e9e actuellement dans le d\u00e9p\u00f4t en ligne de pr\u00e9impression arXiv, les chercheurs notent que les nouveaux mod\u00e8les de langage auto-apprenants du MIT peuvent r\u00e9soudre le probl\u00e8me de la compr\u00e9hension de certaines t\u00e2ches linguistiques que les mod\u00e8les de langage de grande taille rencontrent. Ils qualifient cette avanc\u00e9e r\u00e9volutionnaire d'implication textuelle.<\/span><\/p>\n

Les mod\u00e8les reposent sur le concept selon lequel, s'il y a deux phrases - une pr\u00e9misse et une hypoth\u00e8se, dans le cas o\u00f9 une pr\u00e9misse de la premi\u00e8re phrase est vraie, il est probable que l'hypoth\u00e8se le soit aussi.<\/span><\/p>\n

Dans un<\/span> d\u00e9claration<\/span><\/a> publi\u00e9e sur le blog du MIT CSAIL, un exemple de cette structure serait que si \"tous les chats ont une queue\", l'hypoth\u00e8se \"un chat tabby a une queue\" est susceptible d'\u00eatre vraie. Cette approche permet de r\u00e9duire les biais dans les <\/span>Mod\u00e8les d'IA<\/span><\/a>Selon le communiqu\u00e9, les nouveaux mod\u00e8les linguistiques auto-apprenants du MIT sont plus performants que les mod\u00e8les linguistiques de plus grande taille.<\/span><\/p>\n

\"Nos mod\u00e8les d'implication auto-entra\u00een\u00e9s \u00e0 350 millions de param\u00e8tres, sans \u00e9tiquettes g\u00e9n\u00e9r\u00e9es par l'homme, surpassent les mod\u00e8les de langage supervis\u00e9s avec 137 \u00e0 175 milliards de param\u00e8tres\", a d\u00e9clar\u00e9 Hongyin Luo, associ\u00e9e postdoctorale au MIT CSAIL et auteure principale de l'\u00e9tude. \"<\/span><\/p>\n

Il a \u00e9galement ajout\u00e9 que cette approche pourrait \u00eatre tr\u00e8s b\u00e9n\u00e9fique pour les syst\u00e8mes d'IA actuels et remodeler les syst\u00e8mes d'apprentissage automatique de mani\u00e8re \u00e0 ce qu'ils soient plus \u00e9volutifs, plus fiables et plus rentables lorsqu'ils utilisent des mod\u00e8les linguistiques.<\/span><\/p>\n

Les nouveaux mod\u00e8les d'auto-apprentissage des langues du MIT sont encore limit\u00e9s<\/h2>\n

M\u00eame si les nouveaux mod\u00e8les linguistiques auto-apprenants du MIT promettent beaucoup lorsqu'il s'agit de r\u00e9soudre des probl\u00e8mes de classification binaire, ils restent limit\u00e9s \u00e0 la r\u00e9solution de probl\u00e8mes de classification multi-classes. Cela signifie que l'implication textuelle ne fonctionne pas aussi bien lorsque le mod\u00e8le est confront\u00e9 \u00e0 des choix multiples.<\/span><\/p>\n

Selon James Glass, professeur au MIT et chercheur principal au CSAIL, qui est \u00e9galement l'auteur de l'article, cette recherche pourrait mettre en lumi\u00e8re des m\u00e9thodes efficaces et efficientes pour former les LLM \u00e0 comprendre les probl\u00e8mes d'implication contextuelle.<\/span><\/p>\n

\"Alors que le domaine des LLM conna\u00eet des changements rapides et spectaculaires, cette recherche montre qu'il est possible de produire des mod\u00e8les de langage relativement compacts qui obtiennent de tr\u00e8s bons r\u00e9sultats dans les t\u00e2ches de compr\u00e9hension de r\u00e9f\u00e9rence par rapport \u00e0 leurs homologues de taille \u00e0 peu pr\u00e8s \u00e9quivalente, voire \u00e0 des mod\u00e8les de langage beaucoup plus grands\", a-t-il d\u00e9clar\u00e9.<\/span><\/p>\n

Cette recherche n'est que le d\u00e9but de futures technologies d'IA qui pourraient apprendre par elles-m\u00eames et \u00eatre plus efficaces, durables et ax\u00e9es sur la confidentialit\u00e9 des donn\u00e9es. L'article sur les nouveaux mod\u00e8les linguistiques auto-apprenants du MIT sera pr\u00e9sent\u00e9 en juillet lors de la r\u00e9union de l'Association for Computational Linguistics \u00e0 Toronto. Le projet est \u00e9galement soutenu par le<\/span> Programme Innovation AI de Hong Kong<\/span><\/a>.<\/span><\/p>","protected":false},"excerpt":{"rendered":"

Au cours des six derniers mois, nous avons assist\u00e9 \u00e0 un boom r\u00e9volutionnaire des LLM (Large Language Models) aliment\u00e9s par l'IA. Mais est-il toujours n\u00e9cessaire qu'un produit ou un service d'IA soit bas\u00e9 sur des LLM ? Selon un article, les nouveaux mod\u00e8les linguistiques auto-apprenants du MIT ne sont pas bas\u00e9s sur des LLM et peuvent surpasser certains des autres grands syst\u00e8mes d'IA qui dominent actuellement le secteur. Un groupe de chercheurs du Computer Science and Artificial Intelligence Laboratory (CSAIL) du MIT a mis au point une nouvelle fa\u00e7on d'aborder les mod\u00e8les linguistiques d'IA. Il s'agit d'une r\u00e9alisation r\u00e9volutionnaire qui met l'accent sur des mod\u00e8les de langage plus petits et sur leur capacit\u00e9 \u00e0 r\u00e9pondre aux besoins des utilisateurs.<\/p>","protected":false},"author":4,"featured_media":1280,"comment_status":"closed","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[84],"tags":[118,117],"class_list":["post-1278","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-industry","tag-llms","tag-mit"],"yoast_head":"\nNew MIT self-learning language models can beat other LLMs | DailyAI<\/title>\n<meta name=\"description\" content=\"MIT self-learning language models are not based on LLMs and can outperform some of the other large AI systems that currently lead the industry\" \/>\n<meta name=\"robots\" content=\"index, follow, max-snippet:-1, max-image-preview:large, max-video-preview:-1\" \/>\n<link rel=\"canonical\" href=\"https:\/\/dailyai.com\/fr\/2023\/06\/new-mit-self-learning-language-models-can-beat-other-llms\/\" \/>\n<meta property=\"og:locale\" content=\"fr_FR\" \/>\n<meta property=\"og:type\" content=\"article\" \/>\n<meta property=\"og:title\" content=\"New MIT self-learning language models can beat other LLMs | DailyAI\" \/>\n<meta property=\"og:description\" content=\"MIT self-learning language models are not based on LLMs and can outperform some of the other large AI systems that currently lead the industry\" \/>\n<meta property=\"og:url\" content=\"https:\/\/dailyai.com\/fr\/2023\/06\/new-mit-self-learning-language-models-can-beat-other-llms\/\" \/>\n<meta property=\"og:site_name\" content=\"DailyAI\" \/>\n<meta property=\"article:published_time\" content=\"2023-06-05T17:43:13+00:00\" \/>\n<meta property=\"og:image\" content=\"https:\/\/dailyai.com\/wp-content\/uploads\/2023\/06\/markus-spiske-uPXs5Vx5bIg-unsplash-1-scaled.jpg\" \/>\n\t<meta property=\"og:image:width\" content=\"2560\" \/>\n\t<meta property=\"og:image:height\" content=\"1707\" \/>\n\t<meta property=\"og:image:type\" content=\"image\/jpeg\" \/>\n<meta name=\"author\" content=\"Danica Simic\" \/>\n<meta name=\"twitter:card\" content=\"summary_large_image\" \/>\n<meta name=\"twitter:creator\" content=\"@DailyAIOfficial\" \/>\n<meta name=\"twitter:site\" content=\"@DailyAIOfficial\" \/>\n<meta name=\"twitter:label1\" content=\"\u00c9crit par\" \/>\n\t<meta name=\"twitter:data1\" content=\"Danica Simic\" \/>\n\t<meta name=\"twitter:label2\" content=\"Dur\u00e9e de lecture estim\u00e9e\" \/>\n\t<meta name=\"twitter:data2\" content=\"3 minutes\" \/>\n<script type=\"application\/ld+json\" class=\"yoast-schema-graph\">{\"@context\":\"https:\\\/\\\/schema.org\",\"@graph\":[{\"@type\":\"NewsArticle\",\"@id\":\"https:\\\/\\\/dailyai.com\\\/2023\\\/06\\\/new-mit-self-learning-language-models-can-beat-other-llms\\\/#article\",\"isPartOf\":{\"@id\":\"https:\\\/\\\/dailyai.com\\\/2023\\\/06\\\/new-mit-self-learning-language-models-can-beat-other-llms\\\/\"},\"author\":{\"name\":\"Danica Simic\",\"@id\":\"https:\\\/\\\/dailyai.com\\\/#\\\/schema\\\/person\\\/3ba90fe1f4b0968c7d945612fd92052a\"},\"headline\":\"New MIT self-learning language models can beat other LLMs\",\"datePublished\":\"2023-06-05T17:43:13+00:00\",\"mainEntityOfPage\":{\"@id\":\"https:\\\/\\\/dailyai.com\\\/2023\\\/06\\\/new-mit-self-learning-language-models-can-beat-other-llms\\\/\"},\"wordCount\":623,\"publisher\":{\"@id\":\"https:\\\/\\\/dailyai.com\\\/#organization\"},\"image\":{\"@id\":\"https:\\\/\\\/dailyai.com\\\/2023\\\/06\\\/new-mit-self-learning-language-models-can-beat-other-llms\\\/#primaryimage\"},\"thumbnailUrl\":\"https:\\\/\\\/dailyai.com\\\/wp-content\\\/uploads\\\/2023\\\/06\\\/markus-spiske-uPXs5Vx5bIg-unsplash-1-scaled.jpg\",\"keywords\":[\"LLMS\",\"MIT\"],\"articleSection\":[\"Industry\"],\"inLanguage\":\"fr-FR\"},{\"@type\":\"WebPage\",\"@id\":\"https:\\\/\\\/dailyai.com\\\/2023\\\/06\\\/new-mit-self-learning-language-models-can-beat-other-llms\\\/\",\"url\":\"https:\\\/\\\/dailyai.com\\\/2023\\\/06\\\/new-mit-self-learning-language-models-can-beat-other-llms\\\/\",\"name\":\"New MIT self-learning language models can beat other LLMs | DailyAI\",\"isPartOf\":{\"@id\":\"https:\\\/\\\/dailyai.com\\\/#website\"},\"primaryImageOfPage\":{\"@id\":\"https:\\\/\\\/dailyai.com\\\/2023\\\/06\\\/new-mit-self-learning-language-models-can-beat-other-llms\\\/#primaryimage\"},\"image\":{\"@id\":\"https:\\\/\\\/dailyai.com\\\/2023\\\/06\\\/new-mit-self-learning-language-models-can-beat-other-llms\\\/#primaryimage\"},\"thumbnailUrl\":\"https:\\\/\\\/dailyai.com\\\/wp-content\\\/uploads\\\/2023\\\/06\\\/markus-spiske-uPXs5Vx5bIg-unsplash-1-scaled.jpg\",\"datePublished\":\"2023-06-05T17:43:13+00:00\",\"description\":\"MIT self-learning language models are not based on LLMs and can outperform some of the other large AI systems that currently lead the industry\",\"breadcrumb\":{\"@id\":\"https:\\\/\\\/dailyai.com\\\/2023\\\/06\\\/new-mit-self-learning-language-models-can-beat-other-llms\\\/#breadcrumb\"},\"inLanguage\":\"fr-FR\",\"potentialAction\":[{\"@type\":\"ReadAction\",\"target\":[\"https:\\\/\\\/dailyai.com\\\/2023\\\/06\\\/new-mit-self-learning-language-models-can-beat-other-llms\\\/\"]}]},{\"@type\":\"ImageObject\",\"inLanguage\":\"fr-FR\",\"@id\":\"https:\\\/\\\/dailyai.com\\\/2023\\\/06\\\/new-mit-self-learning-language-models-can-beat-other-llms\\\/#primaryimage\",\"url\":\"https:\\\/\\\/dailyai.com\\\/wp-content\\\/uploads\\\/2023\\\/06\\\/markus-spiske-uPXs5Vx5bIg-unsplash-1-scaled.jpg\",\"contentUrl\":\"https:\\\/\\\/dailyai.com\\\/wp-content\\\/uploads\\\/2023\\\/06\\\/markus-spiske-uPXs5Vx5bIg-unsplash-1-scaled.jpg\",\"width\":5760,\"height\":3840,\"caption\":\"New MIT self-learning language models aren\u2019t based on LLMs (Large Language Models) but can outperform them.\"},{\"@type\":\"BreadcrumbList\",\"@id\":\"https:\\\/\\\/dailyai.com\\\/2023\\\/06\\\/new-mit-self-learning-language-models-can-beat-other-llms\\\/#breadcrumb\",\"itemListElement\":[{\"@type\":\"ListItem\",\"position\":1,\"name\":\"Home\",\"item\":\"https:\\\/\\\/dailyai.com\\\/\"},{\"@type\":\"ListItem\",\"position\":2,\"name\":\"New MIT self-learning language models can beat other LLMs\"}]},{\"@type\":\"WebSite\",\"@id\":\"https:\\\/\\\/dailyai.com\\\/#website\",\"url\":\"https:\\\/\\\/dailyai.com\\\/\",\"name\":\"DailyAI\",\"description\":\"Your Daily Dose of AI News\",\"publisher\":{\"@id\":\"https:\\\/\\\/dailyai.com\\\/#organization\"},\"potentialAction\":[{\"@type\":\"SearchAction\",\"target\":{\"@type\":\"EntryPoint\",\"urlTemplate\":\"https:\\\/\\\/dailyai.com\\\/?s={search_term_string}\"},\"query-input\":{\"@type\":\"PropertyValueSpecification\",\"valueRequired\":true,\"valueName\":\"search_term_string\"}}],\"inLanguage\":\"fr-FR\"},{\"@type\":\"Organization\",\"@id\":\"https:\\\/\\\/dailyai.com\\\/#organization\",\"name\":\"DailyAI\",\"url\":\"https:\\\/\\\/dailyai.com\\\/\",\"logo\":{\"@type\":\"ImageObject\",\"inLanguage\":\"fr-FR\",\"@id\":\"https:\\\/\\\/dailyai.com\\\/#\\\/schema\\\/logo\\\/image\\\/\",\"url\":\"https:\\\/\\\/dailyai.com\\\/wp-content\\\/uploads\\\/2023\\\/06\\\/Daily-Ai_TL_colour.png\",\"contentUrl\":\"https:\\\/\\\/dailyai.com\\\/wp-content\\\/uploads\\\/2023\\\/06\\\/Daily-Ai_TL_colour.png\",\"width\":4501,\"height\":934,\"caption\":\"DailyAI\"},\"image\":{\"@id\":\"https:\\\/\\\/dailyai.com\\\/#\\\/schema\\\/logo\\\/image\\\/\"},\"sameAs\":[\"https:\\\/\\\/x.com\\\/DailyAIOfficial\",\"https:\\\/\\\/www.linkedin.com\\\/company\\\/dailyaiofficial\\\/\",\"https:\\\/\\\/www.youtube.com\\\/@DailyAIOfficial\"]},{\"@type\":\"Person\",\"@id\":\"https:\\\/\\\/dailyai.com\\\/#\\\/schema\\\/person\\\/3ba90fe1f4b0968c7d945612fd92052a\",\"name\":\"Danica Simic\",\"image\":{\"@type\":\"ImageObject\",\"inLanguage\":\"fr-FR\",\"@id\":\"https:\\\/\\\/dailyai.com\\\/wp-content\\\/uploads\\\/2023\\\/06\\\/danica-pp3-96x96.jpeg\",\"url\":\"https:\\\/\\\/dailyai.com\\\/wp-content\\\/uploads\\\/2023\\\/06\\\/danica-pp3-96x96.jpeg\",\"contentUrl\":\"https:\\\/\\\/dailyai.com\\\/wp-content\\\/uploads\\\/2023\\\/06\\\/danica-pp3-96x96.jpeg\",\"caption\":\"Danica Simic\"},\"description\":\"Danica is a data scientist with 7+ years experience in AI & Big Data. She has produced specialist content for the likes of ValueWalk & IntegrateAI and has authored research papers for Singidunum Univeristy.\",\"sameAs\":[\"https:\\\/\\\/www.linkedin.com\\\/in\\\/dasimic\\\/?originalSubdomain=rs\"],\"url\":\"https:\\\/\\\/dailyai.com\\\/fr\\\/author\\\/danicasimic\\\/\"}]}<\/script>\n<!-- \/ Yoast SEO plugin. -->","yoast_head_json":{"title":"Les nouveaux mod\u00e8les linguistiques auto-apprenants du MIT peuvent battre les autres LLM | DailyAI","description":"Les mod\u00e8les linguistiques \u00e0 auto-apprentissage du MIT ne sont pas bas\u00e9s sur les LLM et peuvent surpasser certains des autres grands syst\u00e8mes d'IA qui dominent actuellement le secteur","robots":{"index":"index","follow":"follow","max-snippet":"max-snippet:-1","max-image-preview":"max-image-preview:large","max-video-preview":"max-video-preview:-1"},"canonical":"https:\/\/dailyai.com\/fr\/2023\/06\/new-mit-self-learning-language-models-can-beat-other-llms\/","og_locale":"fr_FR","og_type":"article","og_title":"New MIT self-learning language models can beat other LLMs | DailyAI","og_description":"MIT self-learning language models are not based on LLMs and can outperform some of the other large AI systems that currently lead the industry","og_url":"https:\/\/dailyai.com\/fr\/2023\/06\/new-mit-self-learning-language-models-can-beat-other-llms\/","og_site_name":"DailyAI","article_published_time":"2023-06-05T17:43:13+00:00","og_image":[{"width":2560,"height":1707,"url":"https:\/\/dailyai.com\/wp-content\/uploads\/2023\/06\/markus-spiske-uPXs5Vx5bIg-unsplash-1-scaled.jpg","type":"image\/jpeg"}],"author":"Danica Simic","twitter_card":"summary_large_image","twitter_creator":"@DailyAIOfficial","twitter_site":"@DailyAIOfficial","twitter_misc":{"\u00c9crit par":"Danica Simic","Dur\u00e9e de lecture estim\u00e9e":"3 minutes"},"schema":{"@context":"https:\/\/schema.org","@graph":[{"@type":"NewsArticle","@id":"https:\/\/dailyai.com\/2023\/06\/new-mit-self-learning-language-models-can-beat-other-llms\/#article","isPartOf":{"@id":"https:\/\/dailyai.com\/2023\/06\/new-mit-self-learning-language-models-can-beat-other-llms\/"},"author":{"name":"Danica Simic","@id":"https:\/\/dailyai.com\/#\/schema\/person\/3ba90fe1f4b0968c7d945612fd92052a"},"headline":"New MIT self-learning language models can beat other LLMs","datePublished":"2023-06-05T17:43:13+00:00","mainEntityOfPage":{"@id":"https:\/\/dailyai.com\/2023\/06\/new-mit-self-learning-language-models-can-beat-other-llms\/"},"wordCount":623,"publisher":{"@id":"https:\/\/dailyai.com\/#organization"},"image":{"@id":"https:\/\/dailyai.com\/2023\/06\/new-mit-self-learning-language-models-can-beat-other-llms\/#primaryimage"},"thumbnailUrl":"https:\/\/dailyai.com\/wp-content\/uploads\/2023\/06\/markus-spiske-uPXs5Vx5bIg-unsplash-1-scaled.jpg","keywords":["LLMS","MIT"],"articleSection":["Industry"],"inLanguage":"fr-FR"},{"@type":"WebPage","@id":"https:\/\/dailyai.com\/2023\/06\/new-mit-self-learning-language-models-can-beat-other-llms\/","url":"https:\/\/dailyai.com\/2023\/06\/new-mit-self-learning-language-models-can-beat-other-llms\/","name":"Les nouveaux mod\u00e8les linguistiques auto-apprenants du MIT peuvent battre les autres LLM | DailyAI","isPartOf":{"@id":"https:\/\/dailyai.com\/#website"},"primaryImageOfPage":{"@id":"https:\/\/dailyai.com\/2023\/06\/new-mit-self-learning-language-models-can-beat-other-llms\/#primaryimage"},"image":{"@id":"https:\/\/dailyai.com\/2023\/06\/new-mit-self-learning-language-models-can-beat-other-llms\/#primaryimage"},"thumbnailUrl":"https:\/\/dailyai.com\/wp-content\/uploads\/2023\/06\/markus-spiske-uPXs5Vx5bIg-unsplash-1-scaled.jpg","datePublished":"2023-06-05T17:43:13+00:00","description":"Les mod\u00e8les linguistiques \u00e0 auto-apprentissage du MIT ne sont pas bas\u00e9s sur les LLM et peuvent surpasser certains des autres grands syst\u00e8mes d'IA qui dominent actuellement le secteur","breadcrumb":{"@id":"https:\/\/dailyai.com\/2023\/06\/new-mit-self-learning-language-models-can-beat-other-llms\/#breadcrumb"},"inLanguage":"fr-FR","potentialAction":[{"@type":"ReadAction","target":["https:\/\/dailyai.com\/2023\/06\/new-mit-self-learning-language-models-can-beat-other-llms\/"]}]},{"@type":"ImageObject","inLanguage":"fr-FR","@id":"https:\/\/dailyai.com\/2023\/06\/new-mit-self-learning-language-models-can-beat-other-llms\/#primaryimage","url":"https:\/\/dailyai.com\/wp-content\/uploads\/2023\/06\/markus-spiske-uPXs5Vx5bIg-unsplash-1-scaled.jpg","contentUrl":"https:\/\/dailyai.com\/wp-content\/uploads\/2023\/06\/markus-spiske-uPXs5Vx5bIg-unsplash-1-scaled.jpg","width":5760,"height":3840,"caption":"New MIT self-learning language models aren\u2019t based on LLMs (Large Language Models) but can outperform them."},{"@type":"BreadcrumbList","@id":"https:\/\/dailyai.com\/2023\/06\/new-mit-self-learning-language-models-can-beat-other-llms\/#breadcrumb","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https:\/\/dailyai.com\/"},{"@type":"ListItem","position":2,"name":"New MIT self-learning language models can beat other LLMs"}]},{"@type":"WebSite","@id":"https:\/\/dailyai.com\/#website","url":"https:\/\/dailyai.com\/","name":"DailyAI","description":"Votre dose quotidienne de nouvelles sur l'IA","publisher":{"@id":"https:\/\/dailyai.com\/#organization"},"potentialAction":[{"@type":"SearchAction","target":{"@type":"EntryPoint","urlTemplate":"https:\/\/dailyai.com\/?s={search_term_string}"},"query-input":{"@type":"PropertyValueSpecification","valueRequired":true,"valueName":"search_term_string"}}],"inLanguage":"fr-FR"},{"@type":"Organization","@id":"https:\/\/dailyai.com\/#organization","name":"DailyAI","url":"https:\/\/dailyai.com\/","logo":{"@type":"ImageObject","inLanguage":"fr-FR","@id":"https:\/\/dailyai.com\/#\/schema\/logo\/image\/","url":"https:\/\/dailyai.com\/wp-content\/uploads\/2023\/06\/Daily-Ai_TL_colour.png","contentUrl":"https:\/\/dailyai.com\/wp-content\/uploads\/2023\/06\/Daily-Ai_TL_colour.png","width":4501,"height":934,"caption":"DailyAI"},"image":{"@id":"https:\/\/dailyai.com\/#\/schema\/logo\/image\/"},"sameAs":["https:\/\/x.com\/DailyAIOfficial","https:\/\/www.linkedin.com\/company\/dailyaiofficial\/","https:\/\/www.youtube.com\/@DailyAIOfficial"]},{"@type":"Person","@id":"https:\/\/dailyai.com\/#\/schema\/person\/3ba90fe1f4b0968c7d945612fd92052a","name":"Danica Simic","image":{"@type":"ImageObject","inLanguage":"fr-FR","@id":"https:\/\/dailyai.com\/wp-content\/uploads\/2023\/06\/danica-pp3-96x96.jpeg","url":"https:\/\/dailyai.com\/wp-content\/uploads\/2023\/06\/danica-pp3-96x96.jpeg","contentUrl":"https:\/\/dailyai.com\/wp-content\/uploads\/2023\/06\/danica-pp3-96x96.jpeg","caption":"Danica Simic"},"description":"Danica est une data scientist avec plus de 7 ans d'exp\u00e9rience dans l'IA et le Big Data. Elle a produit du contenu sp\u00e9cialis\u00e9 pour des entreprises telles que ValueWalk et IntegrateAI et a r\u00e9dig\u00e9 des documents de recherche pour l'universit\u00e9 Singidunum.","sameAs":["https:\/\/www.linkedin.com\/in\/dasimic\/?originalSubdomain=rs"],"url":"https:\/\/dailyai.com\/fr\/author\/danicasimic\/"}]}},"_links":{"self":[{"href":"https:\/\/dailyai.com\/fr\/wp-json\/wp\/v2\/posts\/1278","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/dailyai.com\/fr\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/dailyai.com\/fr\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/dailyai.com\/fr\/wp-json\/wp\/v2\/users\/4"}],"replies":[{"embeddable":true,"href":"https:\/\/dailyai.com\/fr\/wp-json\/wp\/v2\/comments?post=1278"}],"version-history":[{"count":3,"href":"https:\/\/dailyai.com\/fr\/wp-json\/wp\/v2\/posts\/1278\/revisions"}],"predecessor-version":[{"id":1376,"href":"https:\/\/dailyai.com\/fr\/wp-json\/wp\/v2\/posts\/1278\/revisions\/1376"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/dailyai.com\/fr\/wp-json\/wp\/v2\/media\/1280"}],"wp:attachment":[{"href":"https:\/\/dailyai.com\/fr\/wp-json\/wp\/v2\/media?parent=1278"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/dailyai.com\/fr\/wp-json\/wp\/v2\/categories?post=1278"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/dailyai.com\/fr\/wp-json\/wp\/v2\/tags?post=1278"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}