{"id":3796,"date":"2020-08-23T06:34:29","date_gmt":"2020-08-23T06:34:29","guid":{"rendered":"https:\/\/www.aiproblog.com\/index.php\/2020\/08\/23\/its-tempting-to-think-that-gp3-will-solve-all-nlp-problems-but-it-does-not\/"},"modified":"2020-08-23T06:34:29","modified_gmt":"2020-08-23T06:34:29","slug":"its-tempting-to-think-that-gp3-will-solve-all-nlp-problems-but-it-does-not","status":"publish","type":"post","link":"https:\/\/www.aiproblog.com\/index.php\/2020\/08\/23\/its-tempting-to-think-that-gp3-will-solve-all-nlp-problems-but-it-does-not\/","title":{"rendered":"It&#8217;s tempting to think that GP3 will solve all NLP problems but it does not"},"content":{"rendered":"<p>Author: ajit jaokar<\/p>\n<div>\n<p>In my previous blog <a href=\"https:\/\/www.datasciencecentral.com\/profiles\/blogs\/what-is-s-driving-the-innovation-in-nlp-and-gpt-3\">what is driving the innovation in nlp and gpt3<\/a> , I talked about how GPT3 has evolved from the basic transformer architecture.<\/p>\n<p>Based on that blog, a start-up approached me saying that they had an idea which they felt could <strong>only<\/strong> be implemented by GPT3.<\/p>\n<p>They were eagerly waiting to be approved (isn&rsquo;t everybody &#8211; he he!)<\/p>\n<p>Apart from waiting for GPT3- there was another critical flaw in their argument<\/p>\n<p>Their idea was not generative i.e. it did not need GPT3 in the first place (or for that matter any similar architecture)<\/p>\n<p><strong>It&rsquo;s tempting to think that GPT-3 will solve all the NLP problems .. but it does not<\/strong><\/p>\n<p>let me explain by this what I mean by this<\/p>\n<p>Below is the basic flow of NLP services and a listing of NLP applications<\/p>\n<p>NLP services include:<\/p>\n<ul>\n<li>Text Summarization<\/li>\n<li>Text Generation<\/li>\n<li>Chatbots<\/li>\n<li>Machine Translation<\/li>\n<li>Text to Speech<\/li>\n<li>Text Classification<\/li>\n<li>Sentence Similarity<\/li>\n<li>Finding similar sentences<\/li>\n<\/ul>\n<p>&nbsp;<a href=\"https:\/\/storage.ning.com\/topology\/rest\/1.0\/file\/get\/7573893666?profile=original\" target=\"_blank\" rel=\"noopener noreferrer\"><img decoding=\"async\" src=\"https:\/\/storage.ning.com\/topology\/rest\/1.0\/file\/get\/7573893666?profile=RESIZE_710x\" class=\"align-full\"><\/a><\/p>\n<p>Image source &ndash; <a href=\"https:\/\/www.linkedin.com\/in\/amitakapoor\/\">Dr Amita Kapoor<\/a><\/p>\n<p>While many of these are generative- not all of them are.<\/p>\n<p>The GPT3 and transformer-based applications basically address the generative elements of NLP<\/p>\n<p>That still leaves a large number of other applications which use NLP but are not generative (for example Text classification or Text summarization).<\/p>\n<p>&nbsp;<\/p>\n<p>You can also look at the same situation from the perspective of <strong>word embeddings<\/strong>. Word embeddings are a type of word representation that allows words with similar meaning to have a similar representation.<\/p>\n<p>Historically, word2vec and GloVe have worked well for word embeddings but these were shallow approaches. Transformers solve this problem by providing a functionality similar to what we see in transfer learning for CNNs (thereby not all layers need to be trained if you use a pre-built model)<\/p>\n<p>&nbsp;<\/p>\n<p><strong>To conclude<\/strong><\/p>\n<p>Hence, we can say that GPT3 is very interesting and will continue to be so.<\/p>\n<p>However, there will be always a subset of NLP applications which will not be covered by any of the transformer-based approaches because they are not generative.<\/p>\n<p>&nbsp;<\/p>\n<\/div>\n<p><a href=\"https:\/\/www.datasciencecentral.com\/xn\/detail\/6448529:BlogPost:978130\">Go to Source<\/a><\/p>\n","protected":false},"excerpt":{"rendered":"<p>Author: ajit jaokar In my previous blog what is driving the innovation in nlp and gpt3 , I talked about how GPT3 has evolved from [&hellip;] <span class=\"read-more-link\"><a class=\"read-more\" href=\"https:\/\/www.aiproblog.com\/index.php\/2020\/08\/23\/its-tempting-to-think-that-gp3-will-solve-all-nlp-problems-but-it-does-not\/\">Read More<\/a><\/span><\/p>\n","protected":false},"author":1,"featured_media":466,"comment_status":"open","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"_bbp_topic_count":0,"_bbp_reply_count":0,"_bbp_total_topic_count":0,"_bbp_total_reply_count":0,"_bbp_voice_count":0,"_bbp_anonymous_reply_count":0,"_bbp_topic_count_hidden":0,"_bbp_reply_count_hidden":0,"_bbp_forum_subforum_count":0,"footnotes":""},"categories":[26],"tags":[],"_links":{"self":[{"href":"https:\/\/www.aiproblog.com\/index.php\/wp-json\/wp\/v2\/posts\/3796"}],"collection":[{"href":"https:\/\/www.aiproblog.com\/index.php\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.aiproblog.com\/index.php\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.aiproblog.com\/index.php\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/www.aiproblog.com\/index.php\/wp-json\/wp\/v2\/comments?post=3796"}],"version-history":[{"count":0,"href":"https:\/\/www.aiproblog.com\/index.php\/wp-json\/wp\/v2\/posts\/3796\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/www.aiproblog.com\/index.php\/wp-json\/wp\/v2\/media\/468"}],"wp:attachment":[{"href":"https:\/\/www.aiproblog.com\/index.php\/wp-json\/wp\/v2\/media?parent=3796"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.aiproblog.com\/index.php\/wp-json\/wp\/v2\/categories?post=3796"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.aiproblog.com\/index.php\/wp-json\/wp\/v2\/tags?post=3796"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}