{"id":593,"date":"2020-06-03T22:28:57","date_gmt":"2020-06-03T16:58:57","guid":{"rendered":"https:\/\/aimonks.com\/aibytes\/?p=593"},"modified":"2020-06-03T22:28:59","modified_gmt":"2020-06-03T16:58:59","slug":"xtreme-a-benchmark-to-spur-development-of-multilingual-ai-models","status":"publish","type":"post","link":"https:\/\/aimonks.com\/aibytes\/2020\/06\/03\/xtreme-a-benchmark-to-spur-development-of-multilingual-ai-models\/","title":{"rendered":"Xtreme- A Benchmark to Spur Development of Multilingual AI Models"},"content":{"rendered":"Reading Time: <\/span> < 1<\/span> minutes<\/span><\/span>\n

Natural language processing<\/strong>, a subdivision of artificial intelligence<\/strong>,  linguistics<\/strong>, information engineering<\/strong>, and computer science<\/strong>. It is concerned with the interactions between human languages and computers, in particular how to program computers to analyze and process large amounts of natural language data. In research, Google <\/strong>found that existing models like multilingual BERT<\/strong>, XML<\/strong>, XML-R<\/strong>, and M4 <\/strong>achieve close to human performance on most existing tasks in English; performance is significantly lower in many other languages<\/strong>.
The release of Google’s NLP <\/strong>systems benchmark Xtreme<\/strong>, consisting of nine tasks<\/strong>, aims to bridge this gap. Researchers believe it can evaluate the AI model’s capabilities to capture knowledge <\/strong>shared across languages so it can be useful for a growing number of natural language applications. The aim is to allow leverage of data<\/strong>–sparse languages <\/strong>to train robust machine learning <\/strong>models.
How does Xtreme work?
\u2981 Diverse languages having coverage of existing tasks and the availability of training data are selected.
\u2981 The models must be pre-trained on multilingual texts to encourage cross-lingual learning.
\u2981 Xtreme evaluates these models with languages that may\/may not have task data available.
\u2981 Then it compares the two and gives a combined score by obtaining zero-shot scores on all tasks.
Since a massive gap between English and other languages remain quite evident, Xtreme could be a step towards catalyzing research in multilingual transfer learning<\/strong>.<\/p>\n\n\n\n

#AIMonks #AI #ArtificialIntelligence #NLP #XML #BERT #XML-4 #M4 #Google #Xtreme #Multilingual #language #Knowledge<\/p>\n","protected":false},"excerpt":{"rendered":"

Reading Time: <\/span> < 1<\/span> minutes<\/span><\/span> Natural language processing, a subdivision of artificial intelligence,  linguistics, information engineering, and computer science. It is concerned with the interactions between human languages and computers, in particular how to program computers to analyze and process large amounts of natural language data. In research, Google found that existing models like multilingual BERT, XML, XML-R, and […]<\/p>\n","protected":false},"author":2,"featured_media":594,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[61],"tags":[35,24,25,566,93,288,85,568,570,88,567,569],"class_list":["post-593","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-digital-transformation","tag-ai","tag-ai-monks","tag-artificial-intelligence","tag-bert","tag-google","tag-knowledge","tag-language","tag-m4","tag-multilingual","tag-nlp","tag-xml-4","tag-xtreme"],"rttpg_featured_image_url":{"full":["https:\/\/aimonks.com\/aibytes\/wp-content\/uploads\/2020\/06\/multilingual.jpg",270,187,false],"landscape":["https:\/\/aimonks.com\/aibytes\/wp-content\/uploads\/2020\/06\/multilingual.jpg",270,187,false],"portraits":["https:\/\/aimonks.com\/aibytes\/wp-content\/uploads\/2020\/06\/multilingual.jpg",270,187,false],"thumbnail":["https:\/\/aimonks.com\/aibytes\/wp-content\/uploads\/2020\/06\/multilingual-150x150.jpg",150,150,true],"medium":["https:\/\/aimonks.com\/aibytes\/wp-content\/uploads\/2020\/06\/multilingual.jpg",270,187,false],"large":["https:\/\/aimonks.com\/aibytes\/wp-content\/uploads\/2020\/06\/multilingual.jpg",270,187,false],"1536x1536":["https:\/\/aimonks.com\/aibytes\/wp-content\/uploads\/2020\/06\/multilingual.jpg",270,187,false],"2048x2048":["https:\/\/aimonks.com\/aibytes\/wp-content\/uploads\/2020\/06\/multilingual.jpg",270,187,false],"hestia-blog":["https:\/\/aimonks.com\/aibytes\/wp-content\/uploads\/2020\/06\/multilingual.jpg",270,187,false]},"rttpg_author":{"display_name":"AI Bytes","author_link":"https:\/\/aimonks.com\/aibytes\/author\/aibytes_kashika\/"},"rttpg_comment":0,"rttpg_category":"Digital Transformation<\/a>","rttpg_excerpt":"Reading Time: < 1 minutes Natural language processing, a subdivision of artificial intelligence,  linguistics, information engineering, and computer science. It is concerned with the interactions between human languages and computers, in particular how to program computers to analyze and process large amounts of natural language data. In research, Google found that existing models like multilingual BERT, XML, XML-R, and…","_links":{"self":[{"href":"https:\/\/aimonks.com\/aibytes\/wp-json\/wp\/v2\/posts\/593","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/aimonks.com\/aibytes\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/aimonks.com\/aibytes\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/aimonks.com\/aibytes\/wp-json\/wp\/v2\/users\/2"}],"replies":[{"embeddable":true,"href":"https:\/\/aimonks.com\/aibytes\/wp-json\/wp\/v2\/comments?post=593"}],"version-history":[{"count":1,"href":"https:\/\/aimonks.com\/aibytes\/wp-json\/wp\/v2\/posts\/593\/revisions"}],"predecessor-version":[{"id":595,"href":"https:\/\/aimonks.com\/aibytes\/wp-json\/wp\/v2\/posts\/593\/revisions\/595"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/aimonks.com\/aibytes\/wp-json\/wp\/v2\/media\/594"}],"wp:attachment":[{"href":"https:\/\/aimonks.com\/aibytes\/wp-json\/wp\/v2\/media?parent=593"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/aimonks.com\/aibytes\/wp-json\/wp\/v2\/categories?post=593"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/aimonks.com\/aibytes\/wp-json\/wp\/v2\/tags?post=593"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}