{"id":8779,"date":"2025-07-20T07:30:59","date_gmt":"2025-07-20T04:30:59","guid":{"rendered":"https:\/\/handoli.com\/index.php\/2025\/07\/20\/data-machina-253\/"},"modified":"2025-07-20T07:30:59","modified_gmt":"2025-07-20T04:30:59","slug":"data-machina-253","status":"publish","type":"post","link":"https:\/\/handoli.com\/index.php\/2025\/07\/20\/data-machina-253\/","title":{"rendered":"Data Machina #253"},"content":{"rendered":"<p><strong>The Google AI Blast . <\/strong>This week OpenAI released a new closed model called GPT-4o (as in omni): <a href=\"https:\/\/openai.com\/index\/hello-gpt-4o\/\">Hello GPT-4o, a model that can reason across audio, vision, and text in real time<\/a>. It seems the model performance in many benchmarks wasn\u2019t as good as many AI pundits expected.<\/p>\n<p>And while many people in the AI community were befuddled and discussing the \u201cflirtatiousness\u201d aspects of GPT-4o, then Google came in and blasted a massive AI storm including SOTA models, new powerful open models, and pretty amazing tools. Here\u2019s my summary on what Google released: <\/p>\n<p><strong>Gemini 1.5 Pro model updates:<\/strong> Lots of improvements in coding, reasoning, translation, multimodality and much more. Some key updates include:<\/p>\n<ul>\n<li>\n<p><a href=\"https:\/\/ai.google.dev\/gemini-api\/docs\/prompting_with_media?lang=python\">multimodal prompting<\/a> to prompt the model with any text, image, audio, and video data<\/p>\n<\/li>\n<li>\n<p><a href=\"https:\/\/ai.google.dev\/gemini-api\/docs\/function-calling\">custom function calling<\/a> to enable real-time interactions with external world<\/p>\n<\/li>\n<li>\n<p><a href=\"https:\/\/ai.google.dev\/gemini-api\/docs\/system-instructions\">systems instructions<\/a> to steer the behaviour of the model based on specific requirements or use cases <\/p>\n<\/li>\n<li>\n<p><a href=\"https:\/\/ai.google.dev\/gemini-api\/docs\/caching\">context caching<\/a> to reduce the cost of requests that contain repeat content with high input token counts<\/p>\n<\/li>\n<li>\n<p>Notably, an extended context size to 2 million tokens! Google researchers say Gemini Pro 1.5 has perfect retrieval (&gt;99%) up to at least 10M tokens, massively beating Claude 3.0 (200k) and GPT-4 Turbo (128k). <\/p>\n<\/li>\n<\/ul>\n<p>If you\u2019re interested to know more read the technical report: <a href=\"https:\/\/storage.googleapis.com\/deepmind-media\/gemini\/gemini_v1_5_report.pdf\">Gemini 1.5: Unlocking multimodal understanding across millions of tokens of context<\/a><\/p>\n<p>Also checkout this video demoing Gemini\u2019s new extended long context<\/p>\n<div class=\"youtube-wrap\" data-attrs='{\"videoId\":\"3N-_lLMDcbs\",\"startTime\":null,\"endTime\":null}' data-component-name=\"Youtube2ToDOM\">\n<div class=\"youtube-inner\"><\/div>\n<\/div>\n<p><strong>New Gemini 1.5 Flash model:<\/strong> The new, smaller <a href=\"https:\/\/deepmind.google\/technologies\/gemini\/flash\/\">Gemini 1.5 Flash model <\/a>is optimised for high-volume, high-frequency tasks at scale, is more cost-efficient to serve. Use it when narrower or high-frequency tasks require fast model\u2019s response and time matters the most. It features a 1 million a long context window. Check this video on Getting started with Gemini Flash.<\/p>\n<div class=\"youtube-wrap\" data-attrs='{\"videoId\":\"ISWNMBY5-o8\",\"startTime\":null,\"endTime\":null}' data-component-name=\"Youtube2ToDOM\">\n<div class=\"youtube-inner\"><\/div>\n<\/div>\n<p><strong>A new, open Vision-Language Model<\/strong>. <a href=\"https:\/\/ai.google.dev\/gemma\/docs\/paligemma\">PaliGemma<\/a> is a powerful open VLM inspired by <a href=\"https:\/\/arxiv.org\/abs\/2310.09199\">PaLI-3 model<\/a>. Built on open components including the SigLIP vision model and the Gemma language model, PaliGemma is designed for class-leading fine-tune performance on a wide range of vision-language tasks. This includes image and short video captioning, visual question answering, understanding text in images, object detection, and object segmentation. Checkout this review by the Hugging Face team: <a href=\"https:\/\/huggingface.co\/blog\/paligemma\">PaliGemma \u2013 Google&#8217;s Cutting-Edge Open Vision Language Model<\/a>.<\/p>\n<div class=\"captioned-image-container\">\n<figure><a class=\"image-link image2 is-viewable-img\" target=\"_blank\" href=\"https:\/\/substackcdn.com\/image\/fetch\/%24s_!3J0a!,f_auto,q_auto:good,fl_progressive:steep\/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F1ef0e227-8c76-4f0c-b4c0-ba8e0a428a05_1012x880.png\" data-component-name=\"Image2ToDOM\">\n<div class=\"image2-inset\"><img decoding=\"async\" src=\"https:\/\/substackcdn.com\/image\/fetch\/%24s_!3J0a!,w_1456,c_limit,f_auto,q_auto:good,fl_progressive:steep\/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F1ef0e227-8c76-4f0c-b4c0-ba8e0a428a05_1012x880.png\" width=\"348\" height=\"302.60869565217394\" data-attrs='{\"src\":\"https:\/\/substack-post-media.s3.amazonaws.com\/public\/images\/1ef0e227-8c76-4f0c-b4c0-ba8e0a428a05_1012x880.png\",\"srcNoWatermark\":null,\"fullscreen\":null,\"imageSize\":null,\"height\":880,\"width\":1012,\"resizeWidth\":348,\"bytes\":108442,\"alt\":null,\"title\":null,\"type\":\"image\/png\",\"href\":null,\"belowTheFold\":true,\"topImage\":false,\"internalRedirect\":null,\"isProcessing\":false,\"align\":null,\"offset\":false}' class=\"sizing-normal\" alt=\"\" loading=\"lazy\"\/>\n<div class=\"image-link-expand\">\n<div class=\"pencraft pc-display-flex pc-gap-8 pc-reset\">\n<div class=\"pencraft pc-reset icon-container restack-image\"><\/div>\n<div class=\"pencraft pc-reset icon-container view-image\"><\/div>\n<\/div>\n<\/div>\n<\/div>\n<p><\/p><\/a><\/figure>\n<\/div>\n<p><strong>Announced new Gemma 2 open  model. <\/strong>A 27 billion parameter model that delivers performance comparable to Mistral and Llama 3 70B at less than half the size. According to Google, this breakthrough efficiency sets a new standard in the open model landscape. <a href=\"https:\/\/techcrunch.com\/2024\/05\/14\/google-announces-gemma-2-a-27b-parameter-version-of-its-open-model-launching-in-june\/\">Gemma 2, a 27B-parameter open model, launching in June<\/a>.<\/p>\n<p><strong>Project Astra: Universal interactive AI Agents<\/strong>. An advanced seeing-and-talking responsive AI agent. The agent uses real-time multi-modality, remembers what it sees and hears to understand context and takes action. It\u2019s also quite proactive, teachable and personal, and has few delays. Checkout this amazing video demo:<\/p>\n<div class=\"youtube-wrap\" data-attrs='{\"videoId\":\"nXVvvRhiGjI\",\"startTime\":\"5\",\"endTime\":null}' data-component-name=\"Youtube2ToDOM\">\n<div class=\"youtube-inner\"><\/div>\n<\/div>\n<p><strong>New ML Model Explorer<\/strong>, a powerful graph visualisation tool that helps one understand, debug, and optimise ML models. It specializes in visualizing large graphs in an intuitive, hierarchical format, but works well for smaller models as well. Blogpost: <a href=\"https:\/\/research.google\/blog\/model-explorer\/\">Model Explorer: Graph visualization for large model development<\/a>.<\/p>\n<div class=\"captioned-image-container\">\n<figure><a class=\"image-link image2 is-viewable-img\" target=\"_blank\" href=\"https:\/\/substackcdn.com\/image\/fetch\/%24s_!yGa4!,f_auto,q_auto:good,fl_progressive:steep\/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F2ab36fc7-1f8b-4f6d-992c-93c65087aa8a_1780x1122.png\" data-component-name=\"Image2ToDOM\">\n<div class=\"image2-inset\"><img decoding=\"async\" src=\"https:\/\/substackcdn.com\/image\/fetch\/%24s_!yGa4!,w_1456,c_limit,f_auto,q_auto:good,fl_progressive:steep\/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F2ab36fc7-1f8b-4f6d-992c-93c65087aa8a_1780x1122.png\" width=\"506\" height=\"319.0302197802198\" data-attrs='{\"src\":\"https:\/\/substack-post-media.s3.amazonaws.com\/public\/images\/2ab36fc7-1f8b-4f6d-992c-93c65087aa8a_1780x1122.png\",\"srcNoWatermark\":null,\"fullscreen\":null,\"imageSize\":null,\"height\":918,\"width\":1456,\"resizeWidth\":506,\"bytes\":null,\"alt\":\"\",\"title\":null,\"type\":null,\"href\":null,\"belowTheFold\":true,\"topImage\":false,\"internalRedirect\":null,\"isProcessing\":false,\"align\":null,\"offset\":false}' class=\"sizing-normal\" alt=\"\" title=\"\" loading=\"lazy\"\/>\n<div class=\"image-link-expand\">\n<div class=\"pencraft pc-display-flex pc-gap-8 pc-reset\">\n<div class=\"pencraft pc-reset icon-container restack-image\"><\/div>\n<div class=\"pencraft pc-reset icon-container view-image\"><\/div>\n<\/div>\n<\/div>\n<\/div>\n<p><\/p><\/a><\/figure>\n<\/div>\n<p><strong>A new AI Safety framework<\/strong>. A set of protocols for proactively identifying future AI capabilities that could cause severe harm and putting in place mechanisms to detect and mitigate them. The focus is on severe risks resulting from powerful capabilities at the model level, such as exceptional agency or sophisticated cyber capabilities. Blogpost: <a href=\"https:\/\/deepmind.google\/discover\/blog\/introducing-the-frontier-safety-framework\/\">Introducing the Frontier Safety Framework<\/a>. <\/p>\n<p><strong>A new Generative AI Toolkit<\/strong> that includes a series of tools to develop and evaluate robust, safe AI apps. It includes an LLM comparator and an interpretability tool. See: <a href=\"https:\/\/ai.google.dev\/responsible\">Responsible Generative AI Toolkit<\/a>. <\/p>\n<p><strong>A new AI Developer competition<\/strong>.  Build an AI App that integrates with Gemini API. Compete for your share of $1 million in cash prizes. Read more about the competition rules, submission guidelines, prizes, and timeline here: <a href=\"https:\/\/ai.google.dev\/competition\">Google Gemini API Developer Competition<\/a>.<\/p>\n<p><strong>Alice\u2019s Adventures in Wonderland reimagined by GenAI<\/strong>.  A set of beautiful interactive stories created by 5 artists using GenAI. Link: <a href=\"https:\/\/infinitewonderland.withgoogle.com\/\">Infinite Wonderland<\/a>.<\/p>\n<div class=\"captioned-image-container\">\n<figure><a class=\"image-link image2\" target=\"_blank\" href=\"https:\/\/substackcdn.com\/image\/fetch\/%24s_!dt-6!,f_auto,q_auto:good,fl_progressive:steep\/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Faa29feee-366c-468a-a6f1-03894a2a1f26_2074x926.png\" data-component-name=\"Image2ToDOM\">\n<div class=\"image2-inset\"><img decoding=\"async\" src=\"https:\/\/substackcdn.com\/image\/fetch\/%24s_!dt-6!,w_1456,c_limit,f_auto,q_auto:good,fl_progressive:steep\/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Faa29feee-366c-468a-a6f1-03894a2a1f26_2074x926.png\" width=\"500\" height=\"223.21428571428572\" data-attrs='{\"src\":\"https:\/\/substack-post-media.s3.amazonaws.com\/public\/images\/aa29feee-366c-468a-a6f1-03894a2a1f26_2074x926.png\",\"srcNoWatermark\":null,\"fullscreen\":null,\"imageSize\":null,\"height\":650,\"width\":1456,\"resizeWidth\":500,\"bytes\":2205175,\"alt\":null,\"title\":null,\"type\":\"image\/png\",\"href\":null,\"belowTheFold\":true,\"topImage\":false,\"internalRedirect\":null,\"isProcessing\":false,\"align\":null,\"offset\":false}' class=\"sizing-normal\" alt=\"\" loading=\"lazy\"\/>\n<div><\/div>\n<\/div>\n<p><\/p><\/a><\/figure>\n<\/div>\n<p>Have a nice week.<\/p>\n<p class=\"button-wrapper\" data-attrs='{\"url\":\"https:\/\/datamachina.substack.com\/subscribe?\",\"text\":\"Subscribe now\",\"action\":null,\"class\":null}' data-component-name=\"ButtonCreateButton\"><a class=\"button primary\" href=\"https:\/\/datamachina.substack.com\/subscribe?\"><span>Subscribe now<\/span><\/a><\/p>\n<h3><strong>10 Link-o-Troned<\/strong><\/h3>\n<ol>\n<li>\n<p><a href=\"https:\/\/stream.thesephist.com\/updates\/1715912318\">On Chat Interfaces &amp; Full Delegation to an AI Agent<\/a><\/p>\n<\/li>\n<li>\n<p><a href=\"https:\/\/phillipi.github.io\/prh\/\">All Neural Nets are Converging to a Platonic Model of Reality<\/a><\/p>\n<\/li>\n<li>\n<p><a href=\"https:\/\/www.sabrina.dev\/p\/test-driving-chatgpt4o-part-4\">Test Driving the AI Capabilities of ChatGPT-4o <\/a><\/p>\n<\/li>\n<li>\n<p><a href=\"https:\/\/www.deeplearning.ai\/short-courses\/multi-ai-agent-systems-with-crewai\/\">[free course] Multi AI Agent Systems with crewAI<\/a><\/p>\n<\/li>\n<li>\n<p><a href=\"https:\/\/www.youtube.com\/watch?v=byH-ARJA4gk\">Embeddings, Transfer Learning &amp; RecSys at Spotify<\/a><\/p>\n<\/li>\n<li>\n<p><a href=\"https:\/\/github.com\/davanstrien\/awesome-synthetic-datasets\">Awesome Synthetic (Text) Datasets<\/a> <\/p>\n<\/li>\n<li>\n<p>\u200a<a href=\"https:\/\/freedium.cfd\/https:\/\/medium.com\/towards-data-science\/n-beats-the-first-interpretable-deep-learning-model-that-worked-for-time-series-forecasting-06920daadac2\">The 1st Interpretable DL Model that Works for Time Series Forecasting<\/a><\/p>\n<\/li>\n<li>\n<p><a href=\"https:\/\/chatqa-project.github.io\/\">NVIDIA ChatQA-1.5 Surpasses GPT-4 on Conversational QA &amp; RAG<\/a><\/p>\n<\/li>\n<li>\n<p><a href=\"https:\/\/chessdream.ai\/?dreamId=4f1434b7-b917-4aec-aaa4-35e1e14c1344\">Chessdream &#8211; A Free AI that Generates Realistic Chess Positions<\/a> <\/p>\n<\/li>\n<li>\n<p><a href=\"https:\/\/www.facebook.com\/thedailyshow\/videos\/new-chatgpt-is-sounding-really-horny\/984712913179450\/\">[comedy] The Daily Show: \u201c<\/a><em><a href=\"https:\/\/www.facebook.com\/thedailyshow\/videos\/new-chatgpt-is-sounding-really-horny\/984712913179450\/\">New GPT-4o is Sounding Really Horny\u201d<\/a><\/em><\/p>\n<\/li>\n<\/ol>\n<div>\n<hr \/>\n<\/div>\n<p class=\"button-wrapper\" data-attrs='{\"url\":\"https:\/\/datamachina.substack.com\",\"text\":\"Share Data Machina with your friends\",\"action\":null,\"class\":\"button-wrapper\"}' data-component-name=\"ButtonCreateButton\"><a class=\"button primary button-wrapper\" href=\"https:\/\/datamachina.substack.com\/\"><span>Share Data Machina with your friends<\/span><\/a><\/p>\n<div>\n<hr \/>\n<\/div>\n<h3><strong>the ML Pythonista<\/strong><\/h3>\n<ol>\n<li>\n<p><a href=\"https:\/\/github.com\/likejazz\/llama3.np\">llama3.np &#8211; Llama 3 Implemented in Pure NumPy<\/a><\/p>\n<\/li>\n<li>\n<p><a href=\"https:\/\/huggingface.co\/parler-tts\/parler-tts-mini-expresso\">Parler-TTS Mini: Expresso &#8211; Natural, Consistent, AI Speech with Emotions<\/a><\/p>\n<\/li>\n<li>\n<p><a href=\"https:\/\/github.com\/Efficient-Large-Model\/VILA\">VILA- An OSS Vision-Language Model for Video &amp; Multi-image Understanding<\/a><\/p>\n<\/li>\n<\/ol>\n<h3><strong>Deep &amp; Other Learning Bits<\/strong><\/h3>\n<ol>\n<li>\n<p><a href=\"https:\/\/www.youtube.com\/watch?v=7zpz_AlFW2w\">KANs Paper Explained &#8211; An Exciting New, DL Paradigm?<\/a> <\/p>\n<\/li>\n<li>\n<p><a href=\"https:\/\/yochan-lab.github.io\/tutorial\/LLMs-Planning\/index.html\">[free tutorial] Planning &amp; Reasoning in LLMs (videos &amp; slides)<\/a><\/p>\n<\/li>\n<li>\n<p><a href=\"https:\/\/www.youtube.com\/watch?v=N6Piou4oYx8\">MAMBA from  Scratch: RNNs are Better and Faster than Transformers<\/a><\/p>\n<\/li>\n<\/ol>\n<h3><strong>AI\/ DL ResearchDocs<\/strong><\/h3>\n<ol>\n<li>\n<p><a href=\"https:\/\/arxiv.org\/abs\/2405.09818\">Meta AI &#8211; Chameleon: Mixed-Modal Early-Fusion FMs that Beat GPT4-V<\/a><\/p>\n<\/li>\n<li>\n<p><a href=\"https:\/\/cat3d.github.io\/\">DeepMind &#8211; CAT3D: Create Anything in 3D with Multi-View Diffusion Models<\/a><\/p>\n<\/li>\n<li>\n<p><a href=\"https:\/\/arxiv.org\/abs\/2405.09673\">DataBricks AI &#8211; LoRA Learns Less and Forgets Less (Underperforms Fine-tuning) <\/a><\/p>\n<\/li>\n<\/ol>\n<h3><strong>MLOps Untangled<\/strong><\/h3>\n<ol>\n<li>\n<p><a href=\"https:\/\/stripe.com\/blog\/shepherd-how-stripe-adapted-chronon-to-scale-ml-feature-development\">Scalable, Next-gen ML Feature Engineering at Stripe <\/a><\/p>\n<\/li>\n<li>\n<p><a href=\"https:\/\/www.youtube.com\/watch?v=jodNnvBFYws\">Streamlining AI Model Deployment &amp; Dynamic Routing<\/a><\/p>\n<\/li>\n<li>\n<p><a href=\"https:\/\/freedium.cfd\/https:\/\/medium.com\/towards-data-science\/common-causes-of-data-leakage-and-how-to-spot-them-17113406f9f8\">Common Causes of ML Models Data Leakage: How to  Deal with Them<\/a><\/p>\n<\/li>\n<\/ol>\n<h3><strong>ML Datasets &amp; Stuff<\/strong><\/h3>\n<ol>\n<li>\n<p><a href=\"https:\/\/ruchitrawal.github.io\/cinepile\/\">CinePile: A Long Video Question Answering Dataset<\/a><\/p>\n<\/li>\n<li>\n<p><a href=\"https:\/\/huggingface.co\/datasets\/Replete-AI\/code_bagel\">Code Bagel: 800 Million Tokens of Unique Coding Data<\/a><\/p>\n<\/li>\n<li>\n<p><a href=\"https:\/\/arxiv.org\/abs\/2405.07425\">Sakuga-42M Dataset: Scaling Up Cartoon Research<\/a><\/p>\n<\/li>\n<\/ol>\n<h3><strong>Postscript, etc <\/strong><\/h3>\n<div class=\"captioned-button-wrap\" data-attrs='{\"url\":\"https:\/\/datamachina.substack.com\/p\/data-machina-253?utm_source=substack&amp;utm_medium=email&amp;utm_content=share&amp;action=share\",\"text\":\"Share\"}' data-component-name=\"CaptionedButtonToDOM\">\n<div class=\"preamble\">\n<p class=\"cta-caption\">Enjoyed this post? Tell your friends about Data Machina. Thanks for reading.<\/p>\n<\/div>\n<p class=\"button-wrapper\" data-attrs='{\"url\":\"https:\/\/datamachina.substack.com\/p\/data-machina-253?utm_source=substack&amp;utm_medium=email&amp;utm_content=share&amp;action=share\",\"text\":\"Share\"}' data-component-name=\"ButtonCreateButton\"><a class=\"button primary\" href=\"https:\/\/datamachina.substack.com\/p\/data-machina-253?utm_source=substack&amp;utm_medium=email&amp;utm_content=share&amp;action=share\"><span>Share<\/span><\/a><\/p>\n<\/div>\n<p>Tips? Suggestions? Feedback?\u00a0<a href=\"mailto:carlos@datamachina.com\">email Carlos<\/a><\/p>\n<p>Curated by\u00a0<a href=\"https:\/\/twitter.com\/ds_ldn\">@ds_ldn\u00a0<\/a>in the middle of the night.<\/p>","protected":false},"excerpt":{"rendered":"<p>The Google AI Blast . This week OpenAI released a new closed model called GPT-4o (as in omni): Hello GPT-4o, a model that can reason across audio, vision, and text in real time. It seems the model performance in many benchmarks wasn\u2019t as good as many AI pundits expected. And while many people in the [&hellip;]<\/p>\n","protected":false},"author":1,"featured_media":8780,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"rop_custom_images_group":[],"rop_custom_messages_group":[],"rop_publish_now":"initial","rop_publish_now_accounts":[],"rop_publish_now_history":[],"rop_publish_now_status":"pending","footnotes":""},"categories":[54],"tags":[],"class_list":["post-8779","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-airepos"],"_links":{"self":[{"href":"https:\/\/handoli.com\/index.php\/wp-json\/wp\/v2\/posts\/8779","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/handoli.com\/index.php\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/handoli.com\/index.php\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/handoli.com\/index.php\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/handoli.com\/index.php\/wp-json\/wp\/v2\/comments?post=8779"}],"version-history":[{"count":0,"href":"https:\/\/handoli.com\/index.php\/wp-json\/wp\/v2\/posts\/8779\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/handoli.com\/index.php\/wp-json\/wp\/v2\/media\/8780"}],"wp:attachment":[{"href":"https:\/\/handoli.com\/index.php\/wp-json\/wp\/v2\/media?parent=8779"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/handoli.com\/index.php\/wp-json\/wp\/v2\/categories?post=8779"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/handoli.com\/index.php\/wp-json\/wp\/v2\/tags?post=8779"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}