{"id":6257,"date":"2024-10-07T14:06:32","date_gmt":"2024-10-07T14:06:32","guid":{"rendered":"https:\/\/www.summetix.com\/llms-leaderboards-long-contexts-und-ihre-limitations\/"},"modified":"2024-10-08T06:38:48","modified_gmt":"2024-10-08T06:38:48","slug":"llms-leaderboards-long-contexts-und-ihre-limitations","status":"publish","type":"post","link":"https:\/\/www.summetix.com\/de\/2024\/10\/07\/llms-leaderboards-long-contexts-und-ihre-limitations\/","title":{"rendered":"LLMs, Leaderboards, Long Contexts und ihre Limitations"},"content":{"rendered":"\n<p>In diesem Beitrag bespricht unser <a href=\"https:\/\/www.linkedin.com\/in\/benjamin-schiller-a695b11a1\">CTO Benjamin Schiller<\/a> einige \u00fcberraschende Ergebnisse unserer neuesten Experimente zu Argument Mining und semantischer Suche. Dieser Beitrag erg\u00e4nzt unsere Reihe zu LLMs und RAG (nach <a href=\"https:\/\/www.summetix.com\/de\/why-do-we-need-arguments-in-rag\/\">Teil 2<\/a> und <a href=\"https:\/\/www.summetix.com\/de\/enhancing-retrieval-augmented-generation-with-argument-mining-a-paradigm-shift-in-ai\/\">Teil 1<\/a>).<\/p>\n\n<figure class=\"wp-block-image\"><a href=\"https:\/\/www.summetix.com\/wp-content\/uploads\/2024\/10\/image.png\"><img decoding=\"async\" src=\"https:\/\/www.summetix.com\/wp-content\/uploads\/2024\/10\/image.png\" alt=\"\"\/><\/a><figcaption class=\"wp-element-caption\">The MTEB leaderboard as of Oct. 2<sup>nd<\/sup> 2024. Source: <a href=\"https:\/\/huggingface.co\/spaces\/mteb\/leaderboard\">Huggingface<\/a><\/figcaption><\/figure>\n\n<p>Wir haben k\u00fcrzlich eine Reihe von Experimenten durchgef\u00fchrt, die von Fortschritten im <a href=\"https:\/\/huggingface.co\/spaces\/mteb\/leaderboard\">MTEB-Leaderboard<\/a> inspiriert wurden. Die Ergebnisse waren \u00fcberraschend. Keines der neueren, state of the art Modelle schnitt bei den dom\u00e4nenspezifischen Aufgaben, mit denen wir sie getestet haben (haupts\u00e4chlich <a href=\"https:\/\/en.wikipedia.org\/wiki\/Argument_mining\">Argument Mining<\/a>), gut ab. Stattdessen sind einige der guten alten Modelle aus dem Jahr 2020 immer noch sehr wettbewerbsf\u00e4hig \u2013 wenn nicht sogar besser.<\/p>\n\n<h2 class=\"wp-block-heading\">Cross-Domain Experimente <\/h2>\n\n<p>Das klingt zwar kontraintuitiv, entspricht aber unseren Erfahrungen aus anderen LLM-Projekten. Dabei ist wichtig zu erw\u00e4hnen, dass unsere Bewertungen streng dom\u00e4nen\u00fcbergreifend sind. Das bedeutet, dass wir immer auf unterschiedlichen Quellen und Themen trainieren und testen. Wenn ein Modell beispielsweise darauf trainiert wird, politische Argumente in sozialen Medien zu identifizieren, testen wir dieses Modell auch auf Produktbewertungen. Modelle, die sich zu sehr auf bestimmte Themen (z. B. im Zusammenhang mit technischen Innovationen wie \u201ek\u00fcnstliche Intelligenz\u201c usw.) oder bestimmte Datentypen (z. B. kurze Beitr\u00e4ge auf X\/Twitter) konzentrieren, werden st\u00e4rker bestraft als solche, die Pr\u00e4zision in einem einzelnen Thema\/Bereich gegen breiteres, bereichs\u00fcbergreifendes Wissen eintauschen. Selbst gr\u00f6\u00dfere Modelle (mehr Parameter, mehr Trainingsdaten) sind nicht unbedingt besser f\u00fcr eine bereichsspezifische Aufgabe geeignet. Unserer Erfahrung nach sagen Fortschritte in den Bestenlisten oft nicht allzu viel \u00fcber unsere internen Anwendungsf\u00e4lle aus. Dies liegt daran, dass sie dazu neigen, sich zu sehr an die Bereiche und Aufgaben der jeweiligen Bestenlisten (in unserem Fall MTEB) anzupassen (zu &#8222;overfitten&#8220;).<\/p>\n\n<h2 class=\"wp-block-heading\">Semantische Suche und Embeddings<\/h2>\n\n<p>Einen \u00e4hnlichen Effekt sehen wir bei der semantischen Suche. Semantische Suche ist stark abh\u00e4ngig von Embeddings ab, und Embeddings sind naturgem\u00e4\u00df sehr dom\u00e4nenabh\u00e4ngig. Neuere Embeddings, die mehr Kontext einbeziehen, scheinen f\u00fcr dieses Problem noch anf\u00e4lliger zu sein. Au\u00dferdem gehen Embeddings mit l\u00e4ngeren Kontexten auf Kosten der Geschwindigkeit und Ressourcen. Komprimierung l\u00f6st dieses Problem zu einem gewissen Grad, geht aber wiederum auf Kosten der Detailgenauigkeit, die f\u00fcr die Suche nach sehr spezifischen Themen erforderlich ist.<sup>[1]<\/sup> Letztendlich ist Standard-BM25 immer noch eine sehr starke Grundlage f\u00fcr viele Anwendungen (siehe auch <a href=\"https:\/\/www.summetix.com\/de\/why-do-we-need-arguments-in-rag\/\">Teil 2 dieser Serie<\/a>).<\/p>\n\n<p>Trotz aller beeindruckenden Fortschritte bei LLMs scheint es bei den meisten realen Anwendungen noch einige Herausforderungen beim Argument Mining und der semantischen Suche zu l\u00f6sen zu geben. Wenn Sie wissen m\u00f6chten, wie SUMMETIX die Herausforderungen der Kundenintelligenz und des Kundensupports mit Argument Mining und LLMs angeht, senden Sie uns eine LinkedIn-Nachricht oder f\u00fcllen Sie das Formular auf <a href=\"https:\/\/www.summetix.com\/de\/\">unserer Website<\/a> aus.<\/p>\n\n<hr class=\"wp-block-separator has-alpha-channel-opacity\"\/>\n\n<p><sup>[1]<\/sup> Wenn Sie sich eingehender mit diesem Thema befassen m\u00f6chten, empfehlen wir Ihnen <a href=\"https:\/\/open.spotify.com\/episode\/3nFGn8YBNvRZRv0MPlFhd1?si=410ae7c50c9746ea\">diese Folge des Podcasts \u201eHow AI is Built\u201c<\/a> mit unserem ehemaligen UKP-Lab-Kollegen Nils Reimers.<\/p>\n\n<hr class=\"wp-block-separator has-alpha-channel-opacity\"\/>\n","protected":false},"excerpt":{"rendered":"<p>In diesem Beitrag bespricht unser CTO Benjamin Schiller einige \u00fcberraschende Ergebnisse unserer neuesten Experimente zu Argument Mining und semantischer Suche. Dieser Beitrag erg\u00e4nzt unsere Reihe zu LLMs und RAG (nach Teil 2 und Teil 1). Wir haben k\u00fcrzlich eine Reihe von Experimenten durchgef\u00fchrt, die von Fortschritten im MTEB-Leaderboard inspiriert wurden. Die Ergebnisse waren \u00fcberraschend. Keines [&hellip;]<\/p>\n","protected":false},"author":7,"featured_media":6243,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"content-type":"","footnotes":""},"categories":[40],"tags":[],"class_list":["post-6257","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-nachrichten"],"yoast_head":"<!-- This site is optimized with the Yoast SEO plugin v27.6 - https:\/\/yoast.com\/product\/yoast-seo-wordpress\/ -->\n<title>Argument mining and semantic search - Limitations<\/title>\n<meta name=\"description\" content=\"Argument mining and semantic search. None of the newer, state-of-the-art, models performed well on the domain-specific tasks on which we tested them.\" \/>\n<meta name=\"robots\" content=\"index, follow, max-snippet:-1, max-image-preview:large, max-video-preview:-1\" \/>\n<link rel=\"canonical\" href=\"https:\/\/www.summetix.com\/de\/2024\/10\/07\/llms-leaderboards-long-contexts-und-ihre-limitations\/\" \/>\n<meta property=\"og:locale\" content=\"de_DE\" \/>\n<meta property=\"og:type\" content=\"article\" \/>\n<meta property=\"og:title\" content=\"Argument mining and semantic search - Limitations\" \/>\n<meta property=\"og:description\" content=\"Argument mining and semantic search. None of the newer, state-of-the-art, models performed well on the domain-specific tasks on which we tested them.\" \/>\n<meta property=\"og:url\" content=\"https:\/\/www.summetix.com\/de\/2024\/10\/07\/llms-leaderboards-long-contexts-und-ihre-limitations\/\" \/>\n<meta property=\"og:site_name\" content=\"summetix\" \/>\n<meta property=\"article:published_time\" content=\"2024-10-07T14:06:32+00:00\" \/>\n<meta property=\"article:modified_time\" content=\"2024-10-08T06:38:48+00:00\" \/>\n<meta property=\"og:image\" content=\"https:\/\/www.summetix.com\/wp-content\/uploads\/2024\/10\/FireflyLLM.jpg\" \/>\n\t<meta property=\"og:image:width\" content=\"908\" \/>\n\t<meta property=\"og:image:height\" content=\"908\" \/>\n\t<meta property=\"og:image:type\" content=\"image\/jpeg\" \/>\n<meta name=\"author\" content=\"Schiller\" \/>\n<meta name=\"twitter:label1\" content=\"Verfasst von\" \/>\n\t<meta name=\"twitter:data1\" content=\"Schiller\" \/>\n\t<meta name=\"twitter:label2\" content=\"Gesch\u00e4tzte Lesezeit\" \/>\n\t<meta name=\"twitter:data2\" content=\"2\u00a0Minuten\" \/>\n<script type=\"application\/ld+json\" class=\"yoast-schema-graph\">{\"@context\":\"https:\\\/\\\/schema.org\",\"@graph\":[{\"@type\":\"Article\",\"@id\":\"https:\\\/\\\/www.summetix.com\\\/de\\\/2024\\\/10\\\/07\\\/llms-leaderboards-long-contexts-und-ihre-limitations\\\/#article\",\"isPartOf\":{\"@id\":\"https:\\\/\\\/www.summetix.com\\\/de\\\/2024\\\/10\\\/07\\\/llms-leaderboards-long-contexts-und-ihre-limitations\\\/\"},\"author\":{\"name\":\"Schiller\",\"@id\":\"https:\\\/\\\/summetix.com\\\/#\\\/schema\\\/person\\\/b21e94c116d19d011c7da0f1afdc5943\"},\"headline\":\"LLMs, Leaderboards, Long Contexts und ihre Limitations\",\"datePublished\":\"2024-10-07T14:06:32+00:00\",\"dateModified\":\"2024-10-08T06:38:48+00:00\",\"mainEntityOfPage\":{\"@id\":\"https:\\\/\\\/www.summetix.com\\\/de\\\/2024\\\/10\\\/07\\\/llms-leaderboards-long-contexts-und-ihre-limitations\\\/\"},\"wordCount\":485,\"commentCount\":0,\"publisher\":{\"@id\":\"https:\\\/\\\/summetix.com\\\/#organization\"},\"image\":{\"@id\":\"https:\\\/\\\/www.summetix.com\\\/de\\\/2024\\\/10\\\/07\\\/llms-leaderboards-long-contexts-und-ihre-limitations\\\/#primaryimage\"},\"thumbnailUrl\":\"https:\\\/\\\/www.summetix.com\\\/wp-content\\\/uploads\\\/2024\\\/10\\\/FireflyLLM.jpg\",\"articleSection\":[\"Nachrichten\"],\"inLanguage\":\"de\",\"potentialAction\":[{\"@type\":\"CommentAction\",\"name\":\"Comment\",\"target\":[\"https:\\\/\\\/www.summetix.com\\\/de\\\/2024\\\/10\\\/07\\\/llms-leaderboards-long-contexts-und-ihre-limitations\\\/#respond\"]}]},{\"@type\":\"WebPage\",\"@id\":\"https:\\\/\\\/www.summetix.com\\\/de\\\/2024\\\/10\\\/07\\\/llms-leaderboards-long-contexts-und-ihre-limitations\\\/\",\"url\":\"https:\\\/\\\/www.summetix.com\\\/de\\\/2024\\\/10\\\/07\\\/llms-leaderboards-long-contexts-und-ihre-limitations\\\/\",\"name\":\"Argument mining and semantic search - Limitations\",\"isPartOf\":{\"@id\":\"https:\\\/\\\/summetix.com\\\/#website\"},\"primaryImageOfPage\":{\"@id\":\"https:\\\/\\\/www.summetix.com\\\/de\\\/2024\\\/10\\\/07\\\/llms-leaderboards-long-contexts-und-ihre-limitations\\\/#primaryimage\"},\"image\":{\"@id\":\"https:\\\/\\\/www.summetix.com\\\/de\\\/2024\\\/10\\\/07\\\/llms-leaderboards-long-contexts-und-ihre-limitations\\\/#primaryimage\"},\"thumbnailUrl\":\"https:\\\/\\\/www.summetix.com\\\/wp-content\\\/uploads\\\/2024\\\/10\\\/FireflyLLM.jpg\",\"datePublished\":\"2024-10-07T14:06:32+00:00\",\"dateModified\":\"2024-10-08T06:38:48+00:00\",\"description\":\"Argument mining and semantic search. None of the newer, state-of-the-art, models performed well on the domain-specific tasks on which we tested them.\",\"breadcrumb\":{\"@id\":\"https:\\\/\\\/www.summetix.com\\\/de\\\/2024\\\/10\\\/07\\\/llms-leaderboards-long-contexts-und-ihre-limitations\\\/#breadcrumb\"},\"inLanguage\":\"de\",\"potentialAction\":[{\"@type\":\"ReadAction\",\"target\":[\"https:\\\/\\\/www.summetix.com\\\/de\\\/2024\\\/10\\\/07\\\/llms-leaderboards-long-contexts-und-ihre-limitations\\\/\"]}]},{\"@type\":\"ImageObject\",\"inLanguage\":\"de\",\"@id\":\"https:\\\/\\\/www.summetix.com\\\/de\\\/2024\\\/10\\\/07\\\/llms-leaderboards-long-contexts-und-ihre-limitations\\\/#primaryimage\",\"url\":\"https:\\\/\\\/www.summetix.com\\\/wp-content\\\/uploads\\\/2024\\\/10\\\/FireflyLLM.jpg\",\"contentUrl\":\"https:\\\/\\\/www.summetix.com\\\/wp-content\\\/uploads\\\/2024\\\/10\\\/FireflyLLM.jpg\",\"width\":908,\"height\":908},{\"@type\":\"BreadcrumbList\",\"@id\":\"https:\\\/\\\/www.summetix.com\\\/de\\\/2024\\\/10\\\/07\\\/llms-leaderboards-long-contexts-und-ihre-limitations\\\/#breadcrumb\",\"itemListElement\":[{\"@type\":\"ListItem\",\"position\":1,\"name\":\"Home\",\"item\":\"https:\\\/\\\/summetix.com\\\/de\\\/\"},{\"@type\":\"ListItem\",\"position\":2,\"name\":\"LLMs, Leaderboards, Long Contexts und ihre Limitations\"}]},{\"@type\":\"WebSite\",\"@id\":\"https:\\\/\\\/summetix.com\\\/#website\",\"url\":\"https:\\\/\\\/summetix.com\\\/\",\"name\":\"summetix\",\"description\":\"Discover hidden insights in customer feedback and complex qualitative data. summetix uses proprietary Argument Mining and large language models to discover patterns and trends that help your business save and make money.\",\"publisher\":{\"@id\":\"https:\\\/\\\/summetix.com\\\/#organization\"},\"potentialAction\":[{\"@type\":\"SearchAction\",\"target\":{\"@type\":\"EntryPoint\",\"urlTemplate\":\"https:\\\/\\\/summetix.com\\\/?s={search_term_string}\"},\"query-input\":{\"@type\":\"PropertyValueSpecification\",\"valueRequired\":true,\"valueName\":\"search_term_string\"}}],\"inLanguage\":\"de\"},{\"@type\":\"Organization\",\"@id\":\"https:\\\/\\\/summetix.com\\\/#organization\",\"name\":\"summetix\",\"url\":\"https:\\\/\\\/summetix.com\\\/\",\"logo\":{\"@type\":\"ImageObject\",\"inLanguage\":\"de\",\"@id\":\"https:\\\/\\\/summetix.com\\\/#\\\/schema\\\/logo\\\/image\\\/\",\"url\":\"https:\\\/\\\/www.summetix.com\\\/wp-content\\\/uploads\\\/2023\\\/12\\\/summetix_smooth_text.png\",\"contentUrl\":\"https:\\\/\\\/www.summetix.com\\\/wp-content\\\/uploads\\\/2023\\\/12\\\/summetix_smooth_text.png\",\"width\":2000,\"height\":304,\"caption\":\"summetix\"},\"image\":{\"@id\":\"https:\\\/\\\/summetix.com\\\/#\\\/schema\\\/logo\\\/image\\\/\"}},{\"@type\":\"Person\",\"@id\":\"https:\\\/\\\/summetix.com\\\/#\\\/schema\\\/person\\\/b21e94c116d19d011c7da0f1afdc5943\",\"name\":\"Schiller\",\"image\":{\"@type\":\"ImageObject\",\"inLanguage\":\"de\",\"@id\":\"https:\\\/\\\/secure.gravatar.com\\\/avatar\\\/40b316d6cab436ceaf4711867a2248236aa5cde559326a08790bfe23ba83d85b?s=96&d=mm&r=g\",\"url\":\"https:\\\/\\\/secure.gravatar.com\\\/avatar\\\/40b316d6cab436ceaf4711867a2248236aa5cde559326a08790bfe23ba83d85b?s=96&d=mm&r=g\",\"contentUrl\":\"https:\\\/\\\/secure.gravatar.com\\\/avatar\\\/40b316d6cab436ceaf4711867a2248236aa5cde559326a08790bfe23ba83d85b?s=96&d=mm&r=g\",\"caption\":\"Schiller\"},\"url\":\"https:\\\/\\\/www.summetix.com\\\/de\\\/author\\\/schiller\\\/\"}]}<\/script>\n<!-- \/ Yoast SEO plugin. -->","yoast_head_json":{"title":"Argument mining and semantic search - Limitations","description":"Argument mining and semantic search. None of the newer, state-of-the-art, models performed well on the domain-specific tasks on which we tested them.","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:\/\/www.summetix.com\/de\/2024\/10\/07\/llms-leaderboards-long-contexts-und-ihre-limitations\/","og_locale":"de_DE","og_type":"article","og_title":"Argument mining and semantic search - Limitations","og_description":"Argument mining and semantic search. None of the newer, state-of-the-art, models performed well on the domain-specific tasks on which we tested them.","og_url":"https:\/\/www.summetix.com\/de\/2024\/10\/07\/llms-leaderboards-long-contexts-und-ihre-limitations\/","og_site_name":"summetix","article_published_time":"2024-10-07T14:06:32+00:00","article_modified_time":"2024-10-08T06:38:48+00:00","og_image":[{"width":908,"height":908,"url":"https:\/\/www.summetix.com\/wp-content\/uploads\/2024\/10\/FireflyLLM.jpg","type":"image\/jpeg"}],"author":"Schiller","twitter_misc":{"Verfasst von":"Schiller","Gesch\u00e4tzte Lesezeit":"2\u00a0Minuten"},"schema":{"@context":"https:\/\/schema.org","@graph":[{"@type":"Article","@id":"https:\/\/www.summetix.com\/de\/2024\/10\/07\/llms-leaderboards-long-contexts-und-ihre-limitations\/#article","isPartOf":{"@id":"https:\/\/www.summetix.com\/de\/2024\/10\/07\/llms-leaderboards-long-contexts-und-ihre-limitations\/"},"author":{"name":"Schiller","@id":"https:\/\/summetix.com\/#\/schema\/person\/b21e94c116d19d011c7da0f1afdc5943"},"headline":"LLMs, Leaderboards, Long Contexts und ihre Limitations","datePublished":"2024-10-07T14:06:32+00:00","dateModified":"2024-10-08T06:38:48+00:00","mainEntityOfPage":{"@id":"https:\/\/www.summetix.com\/de\/2024\/10\/07\/llms-leaderboards-long-contexts-und-ihre-limitations\/"},"wordCount":485,"commentCount":0,"publisher":{"@id":"https:\/\/summetix.com\/#organization"},"image":{"@id":"https:\/\/www.summetix.com\/de\/2024\/10\/07\/llms-leaderboards-long-contexts-und-ihre-limitations\/#primaryimage"},"thumbnailUrl":"https:\/\/www.summetix.com\/wp-content\/uploads\/2024\/10\/FireflyLLM.jpg","articleSection":["Nachrichten"],"inLanguage":"de","potentialAction":[{"@type":"CommentAction","name":"Comment","target":["https:\/\/www.summetix.com\/de\/2024\/10\/07\/llms-leaderboards-long-contexts-und-ihre-limitations\/#respond"]}]},{"@type":"WebPage","@id":"https:\/\/www.summetix.com\/de\/2024\/10\/07\/llms-leaderboards-long-contexts-und-ihre-limitations\/","url":"https:\/\/www.summetix.com\/de\/2024\/10\/07\/llms-leaderboards-long-contexts-und-ihre-limitations\/","name":"Argument mining and semantic search - Limitations","isPartOf":{"@id":"https:\/\/summetix.com\/#website"},"primaryImageOfPage":{"@id":"https:\/\/www.summetix.com\/de\/2024\/10\/07\/llms-leaderboards-long-contexts-und-ihre-limitations\/#primaryimage"},"image":{"@id":"https:\/\/www.summetix.com\/de\/2024\/10\/07\/llms-leaderboards-long-contexts-und-ihre-limitations\/#primaryimage"},"thumbnailUrl":"https:\/\/www.summetix.com\/wp-content\/uploads\/2024\/10\/FireflyLLM.jpg","datePublished":"2024-10-07T14:06:32+00:00","dateModified":"2024-10-08T06:38:48+00:00","description":"Argument mining and semantic search. None of the newer, state-of-the-art, models performed well on the domain-specific tasks on which we tested them.","breadcrumb":{"@id":"https:\/\/www.summetix.com\/de\/2024\/10\/07\/llms-leaderboards-long-contexts-und-ihre-limitations\/#breadcrumb"},"inLanguage":"de","potentialAction":[{"@type":"ReadAction","target":["https:\/\/www.summetix.com\/de\/2024\/10\/07\/llms-leaderboards-long-contexts-und-ihre-limitations\/"]}]},{"@type":"ImageObject","inLanguage":"de","@id":"https:\/\/www.summetix.com\/de\/2024\/10\/07\/llms-leaderboards-long-contexts-und-ihre-limitations\/#primaryimage","url":"https:\/\/www.summetix.com\/wp-content\/uploads\/2024\/10\/FireflyLLM.jpg","contentUrl":"https:\/\/www.summetix.com\/wp-content\/uploads\/2024\/10\/FireflyLLM.jpg","width":908,"height":908},{"@type":"BreadcrumbList","@id":"https:\/\/www.summetix.com\/de\/2024\/10\/07\/llms-leaderboards-long-contexts-und-ihre-limitations\/#breadcrumb","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https:\/\/summetix.com\/de\/"},{"@type":"ListItem","position":2,"name":"LLMs, Leaderboards, Long Contexts und ihre Limitations"}]},{"@type":"WebSite","@id":"https:\/\/summetix.com\/#website","url":"https:\/\/summetix.com\/","name":"summetix","description":"Discover hidden insights in customer feedback and complex qualitative data. summetix uses proprietary Argument Mining and large language models to discover patterns and trends that help your business save and make money.","publisher":{"@id":"https:\/\/summetix.com\/#organization"},"potentialAction":[{"@type":"SearchAction","target":{"@type":"EntryPoint","urlTemplate":"https:\/\/summetix.com\/?s={search_term_string}"},"query-input":{"@type":"PropertyValueSpecification","valueRequired":true,"valueName":"search_term_string"}}],"inLanguage":"de"},{"@type":"Organization","@id":"https:\/\/summetix.com\/#organization","name":"summetix","url":"https:\/\/summetix.com\/","logo":{"@type":"ImageObject","inLanguage":"de","@id":"https:\/\/summetix.com\/#\/schema\/logo\/image\/","url":"https:\/\/www.summetix.com\/wp-content\/uploads\/2023\/12\/summetix_smooth_text.png","contentUrl":"https:\/\/www.summetix.com\/wp-content\/uploads\/2023\/12\/summetix_smooth_text.png","width":2000,"height":304,"caption":"summetix"},"image":{"@id":"https:\/\/summetix.com\/#\/schema\/logo\/image\/"}},{"@type":"Person","@id":"https:\/\/summetix.com\/#\/schema\/person\/b21e94c116d19d011c7da0f1afdc5943","name":"Schiller","image":{"@type":"ImageObject","inLanguage":"de","@id":"https:\/\/secure.gravatar.com\/avatar\/40b316d6cab436ceaf4711867a2248236aa5cde559326a08790bfe23ba83d85b?s=96&d=mm&r=g","url":"https:\/\/secure.gravatar.com\/avatar\/40b316d6cab436ceaf4711867a2248236aa5cde559326a08790bfe23ba83d85b?s=96&d=mm&r=g","contentUrl":"https:\/\/secure.gravatar.com\/avatar\/40b316d6cab436ceaf4711867a2248236aa5cde559326a08790bfe23ba83d85b?s=96&d=mm&r=g","caption":"Schiller"},"url":"https:\/\/www.summetix.com\/de\/author\/schiller\/"}]}},"_links":{"self":[{"href":"https:\/\/www.summetix.com\/de\/wp-json\/wp\/v2\/posts\/6257","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.summetix.com\/de\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.summetix.com\/de\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.summetix.com\/de\/wp-json\/wp\/v2\/users\/7"}],"replies":[{"embeddable":true,"href":"https:\/\/www.summetix.com\/de\/wp-json\/wp\/v2\/comments?post=6257"}],"version-history":[{"count":2,"href":"https:\/\/www.summetix.com\/de\/wp-json\/wp\/v2\/posts\/6257\/revisions"}],"predecessor-version":[{"id":6262,"href":"https:\/\/www.summetix.com\/de\/wp-json\/wp\/v2\/posts\/6257\/revisions\/6262"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/www.summetix.com\/de\/wp-json\/wp\/v2\/media\/6243"}],"wp:attachment":[{"href":"https:\/\/www.summetix.com\/de\/wp-json\/wp\/v2\/media?parent=6257"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.summetix.com\/de\/wp-json\/wp\/v2\/categories?post=6257"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.summetix.com\/de\/wp-json\/wp\/v2\/tags?post=6257"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}