{"id":1110,"date":"2024-12-06T16:17:30","date_gmt":"2024-12-06T09:17:30","guid":{"rendered":"https:\/\/mina.ai.vn\/?p=1110"},"modified":"2024-12-06T16:17:30","modified_gmt":"2024-12-06T09:17:30","slug":"a-peek-inside-large-language-modelsmind-real-time-sentence-construction-visualized","status":"publish","type":"post","link":"http:\/\/mina.id.vn\/?p=1110","title":{"rendered":"A peek inside Large Language Model\u2019s mind: Real-time sentence construction visualized"},"content":{"rendered":"\n<p>Data scientist Santiago Ortiz has created a captivating visualization of how Large Language Model constructs sentences step by step. Using the prompt <strong>\u201cIntelligence is\u201d<\/strong> and setting a high temperature of 1.6 for diverse responses, Ortiz ran hundreds of iterations to map the AI\u2019s thought process.<\/p>\n\n\n\n<p>To make the intricate data comprehensible, Ortiz utilized <strong>Principal Component Analysis (PCA)<\/strong> to reduce 1536-dimensional word embeddings into a 3D space, showing how responses unfold word by word.<\/p>\n\n\n\n<p>The project features two interconnected visualizations:<\/p>\n\n\n\n<p>&nbsp;&nbsp;&nbsp;&nbsp; \u2022&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; <strong>A 3D cube<\/strong> with branching paths representing the evolution of text sequences.<\/p>\n\n\n\n<p>&nbsp;&nbsp;&nbsp;&nbsp; \u2022&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; <strong>A tree diagram<\/strong> that highlights word probabilities for each path.<\/p>\n\n\n\n<p>By hovering over the points, users can observe specific sentence trajectories, revealing the decision-making patterns of LLM in real-time.<\/p>\n\n\n\n<p>Experience the project and gain a unique insight into AI\u2019s thought process at <a href=\"https:\/\/moebio.com\/mind\" target=\"_blank\" rel=\"noreferrer noopener\">Moebio\u2019s Mind<\/a>.<\/p>\n","protected":false},"excerpt":{"rendered":"<p>Data scientist Santiago Ortiz has created a captivating visualization of how Large Language Model constructs sentences step by step. Using the prompt \u201cIntelligence is\u201d and setting a high temperature of 1.6 for diverse responses, Ortiz ran hundreds of iterations to map the AI\u2019s thought process. To make the intricate data comprehensible, Ortiz utilized Principal Component [&hellip;]<\/p>\n","protected":false},"author":1,"featured_media":1112,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"pagelayer_contact_templates":[],"_pagelayer_content":"","footnotes":""},"categories":[3],"tags":[11,33,45,122,194],"class_list":["post-1110","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-knowledge","tag-ai","tag-artificial-intelligence","tag-chatgpt","tag-llm","tag-technology"],"_links":{"self":[{"href":"http:\/\/mina.id.vn\/index.php?rest_route=\/wp\/v2\/posts\/1110","targetHints":{"allow":["GET"]}}],"collection":[{"href":"http:\/\/mina.id.vn\/index.php?rest_route=\/wp\/v2\/posts"}],"about":[{"href":"http:\/\/mina.id.vn\/index.php?rest_route=\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"http:\/\/mina.id.vn\/index.php?rest_route=\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"http:\/\/mina.id.vn\/index.php?rest_route=%2Fwp%2Fv2%2Fcomments&post=1110"}],"version-history":[{"count":0,"href":"http:\/\/mina.id.vn\/index.php?rest_route=\/wp\/v2\/posts\/1110\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"http:\/\/mina.id.vn\/index.php?rest_route=\/wp\/v2\/media\/1112"}],"wp:attachment":[{"href":"http:\/\/mina.id.vn\/index.php?rest_route=%2Fwp%2Fv2%2Fmedia&parent=1110"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"http:\/\/mina.id.vn\/index.php?rest_route=%2Fwp%2Fv2%2Fcategories&post=1110"},{"taxonomy":"post_tag","embeddable":true,"href":"http:\/\/mina.id.vn\/index.php?rest_route=%2Fwp%2Fv2%2Ftags&post=1110"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}