Fosstodon

Alexander DunkelNew article about <a href="https://himself.alexanderdunkel.com/tags/embedding" class="mention hashtag" rel="nofollow noopener" target="_blank">#embedding</a> based multilingual search [1]. Dominik Weckmüller, the lead author, contributed the core of this article, which focuses on workflow, cartography and social media data. This is a great read if you are working in fields such as <a href="https://himself.alexanderdunkel.com/tags/NLP" class="mention hashtag" rel="nofollow noopener" target="_blank">#NLP</a> and <a href="https://himself.alexanderdunkel.com/tags/cartography" class="mention hashtag" rel="nofollow noopener" target="_blank">#cartography</a>.[1]: <a href="https://doi.org/10.1007/s41651-025-00232-5" rel="nofollow noopener" translate="no" target="_blank">https://doi.org/10.1007/s41651-025-00232-5</a>

Inautilo<a href="https://mastodon.social/tags/Development" class="mention hashtag" rel="nofollow noopener" target="_blank">#Development</a> <a href="https://mastodon.social/tags/Approaches" class="mention hashtag" rel="nofollow noopener" target="_blank">#Approaches</a> Responsive video is (almost) easy now · Ready to support both horizontal and vertical video? <a href="https://ilo.im/165isi" rel="nofollow noopener" translate="no" target="_blank">https://ilo.im/165isi</a>_____ <a href="https://mastodon.social/tags/Embedding" class="mention hashtag" rel="nofollow noopener" target="_blank">#Embedding</a> <a href="https://mastodon.social/tags/Video" class="mention hashtag" rel="nofollow noopener" target="_blank">#Video</a> <a href="https://mastodon.social/tags/MP4" class="mention hashtag" rel="nofollow noopener" target="_blank">#MP4</a> <a href="https://mastodon.social/tags/Layout" class="mention hashtag" rel="nofollow noopener" target="_blank">#Layout</a> <a href="https://mastodon.social/tags/Content" class="mention hashtag" rel="nofollow noopener" target="_blank">#Content</a> <a href="https://mastodon.social/tags/WebDesign" class="mention hashtag" rel="nofollow noopener" target="_blank">#WebDesign</a> <a href="https://mastodon.social/tags/WebDev" class="mention hashtag" rel="nofollow noopener" target="_blank">#WebDev</a> <a href="https://mastodon.social/tags/Frontend" class="mention hashtag" rel="nofollow noopener" target="_blank">#Frontend</a> <a href="https://mastodon.social/tags/CSS" class="mention hashtag" rel="nofollow noopener" target="_blank">#CSS</a> <a href="https://mastodon.social/tags/JavaScript" class="mention hashtag" rel="nofollow noopener" target="_blank">#JavaScript</a>

Alessio Pomaro🧠 Il nuovo modello <a href="https://mastodon.uno/tags/Gemini" class="mention hashtag" rel="nofollow noopener" target="_blank">#Gemini</a> <a href="https://mastodon.uno/tags/Embedding" class="mention hashtag" rel="nofollow noopener" target="_blank">#Embedding</a> (gemini-embedding-001) è ora disponibile pubblicamente tramite l’API Gemini e Vertex AI.  👉 Per approfondire: <a href="https://www.linkedin.com/posts/alessiopomaro_gemini-embedding-ai-activity-7353012458518626306-XL4y" rel="nofollow noopener" translate="no" target="_blank">https://www.linkedin.com/posts/alessiopomaro_gemini-embedding-ai-activity-7353012458518626306-XL4y</a>___  ✉️ 𝗦𝗲 𝘃𝘂𝗼𝗶 𝗿𝗶𝗺𝗮𝗻𝗲𝗿𝗲 𝗮𝗴𝗴𝗶𝗼𝗿𝗻𝗮𝘁𝗼/𝗮 𝘀𝘂 𝗾𝘂𝗲𝘀𝘁𝗲 𝘁𝗲𝗺𝗮𝘁𝗶𝗰𝗵𝗲, 𝗶𝘀𝗰𝗿𝗶𝘃𝗶𝘁𝗶 𝗮𝗹𝗹𝗮 𝗺𝗶𝗮 𝗻𝗲𝘄𝘀𝗹𝗲𝘁𝘁𝗲𝗿: <a href="https://bit.ly/newsletter-alessiopomaro" rel="nofollow noopener" translate="no" target="_blank">https://bit.ly/newsletter-alessiopomaro</a> <a href="https://mastodon.uno/tags/AI" class="mention hashtag" rel="nofollow noopener" target="_blank">#AI</a> <a href="https://mastodon.uno/tags/GenAI" class="mention hashtag" rel="nofollow noopener" target="_blank">#GenAI</a> <a href="https://mastodon.uno/tags/GenerativeAI" class="mention hashtag" rel="nofollow noopener" target="_blank">#GenerativeAI</a> <a href="https://mastodon.uno/tags/IntelligenzaArtificiale" class="mention hashtag" rel="nofollow noopener" target="_blank">#IntelligenzaArtificiale</a> <a href="https://mastodon.uno/tags/LLM" class="mention hashtag" rel="nofollow noopener" target="_blank">#LLM</a>

Hacker NewsEmbedding User-Defined Indexes in Apache Parquet<a href="https://datafusion.apache.org/blog/2025/07/14/user-defined-parquet-indexes/" rel="nofollow noopener" translate="no" target="_blank">https://datafusion.apache.org/blog/2025/07/14/user-defined-parquet-indexes/</a><a href="https://mastodon.social/tags/HackerNews" class="mention hashtag" rel="nofollow noopener" target="_blank">#HackerNews</a> <a href="https://mastodon.social/tags/Embedding" class="mention hashtag" rel="nofollow noopener" target="_blank">#Embedding</a> <a href="https://mastodon.social/tags/User" class="mention hashtag" rel="nofollow noopener" target="_blank">#User</a>-Defined <a href="https://mastodon.social/tags/Indexes" class="mention hashtag" rel="nofollow noopener" target="_blank">#Indexes</a> <a href="https://mastodon.social/tags/in" class="mention hashtag" rel="nofollow noopener" target="_blank">#in</a> <a href="https://mastodon.social/tags/Apache" class="mention hashtag" rel="nofollow noopener" target="_blank">#Apache</a> <a href="https://mastodon.social/tags/Parquet" class="mention hashtag" rel="nofollow noopener" target="_blank">#Parquet</a> <a href="https://mastodon.social/tags/ApacheParquet" class="mention hashtag" rel="nofollow noopener" target="_blank">#ApacheParquet</a> <a href="https://mastodon.social/tags/UserDefinedIndexes" class="mention hashtag" rel="nofollow noopener" target="_blank">#UserDefinedIndexes</a> <a href="https://mastodon.social/tags/DataFusion" class="mention hashtag" rel="nofollow noopener" target="_blank">#DataFusion</a> <a href="https://mastodon.social/tags/BigData" class="mention hashtag" rel="nofollow noopener" target="_blank">#BigData</a> <a href="https://mastodon.social/tags/Analytics" class="mention hashtag" rel="nofollow noopener" target="_blank">#Analytics</a>

:rss: DevelopersIOAWS MemoryDB のセマンティックキャッシュによる LLM アプリケーションの高速化検証 <a href="https://dev.classmethod.jp/articles/aws-memorydb-semantic-cache-llm-performance/" rel="nofollow noopener" translate="no" target="_blank">https://dev.classmethod.jp/articles/aws-memorydb-semantic-cache-llm-performance/</a><a href="https://rss-mstdn.studiofreesia.com/tags/dev_classmethod" class="mention hashtag" rel="nofollow noopener" target="_blank">#dev_classmethod</a> <a href="https://rss-mstdn.studiofreesia.com/tags/Amazon_MemoryDB" class="mention hashtag" rel="nofollow noopener" target="_blank">#Amazon_MemoryDB</a> <a href="https://rss-mstdn.studiofreesia.com/tags/embedding" class="mention hashtag" rel="nofollow noopener" target="_blank">#embedding</a> <a href="https://rss-mstdn.studiofreesia.com/tags/Amazon_Bedrock" class="mention hashtag" rel="nofollow noopener" target="_blank">#Amazon_Bedrock</a> <a href="https://rss-mstdn.studiofreesia.com/tags/%E7%94%9F%E6%88%90AI" class="mention hashtag" rel="nofollow noopener" target="_blank">#生成AI</a> <a href="https://rss-mstdn.studiofreesia.com/tags/Claude" class="mention hashtag" rel="nofollow noopener" target="_blank">#Claude</a>

N-gated Hacker News🔮✨ Behold the latest <a href="https://mastodon.social/tags/revolution" class="mention hashtag" rel="nofollow noopener" target="_blank">#revolution</a> in web tech: <a href="https://mastodon.social/tags/embedding" class="mention hashtag" rel="nofollow noopener" target="_blank">#embedding</a> Lisp—because what the internet truly needs is more parentheses! 🌐🙃 A bold step forward for those who believe <a href="https://mastodon.social/tags/JavaScript" class="mention hashtag" rel="nofollow noopener" target="_blank">#JavaScript</a> is just too mainstream and not nearly cryptic enough. 🎩🧐 <a href="https://turtleware.eu/static/paste/wecl-test-gl/main.html" rel="nofollow noopener" translate="no" target="_blank">https://turtleware.eu/static/paste/wecl-test-gl/main.html</a> <a href="https://mastodon.social/tags/webtech" class="mention hashtag" rel="nofollow noopener" target="_blank">#webtech</a> <a href="https://mastodon.social/tags/Lisp" class="mention hashtag" rel="nofollow noopener" target="_blank">#Lisp</a> <a href="https://mastodon.social/tags/alternatives" class="mention hashtag" rel="nofollow noopener" target="_blank">#alternatives</a> <a href="https://mastodon.social/tags/coding" class="mention hashtag" rel="nofollow noopener" target="_blank">#coding</a> <a href="https://mastodon.social/tags/humor" class="mention hashtag" rel="nofollow noopener" target="_blank">#humor</a> <a href="https://mastodon.social/tags/HackerNews" class="mention hashtag" rel="nofollow noopener" target="_blank">#HackerNews</a> <a href="https://mastodon.social/tags/ngated" class="mention hashtag" rel="nofollow noopener" target="_blank">#ngated</a>

Rost GlukhovReranking text documents with Ollama and Qwen3 Embedding model - in Golang: <a href="https://www.glukhov.org/post/2025/06/reranking-with-ollama-qwen3-embedding-golang/" rel="nofollow noopener" translate="no" target="_blank">https://www.glukhov.org/post/2025/06/reranking-with-ollama-qwen3-embedding-golang/</a> <a href="https://techhub.social/tags/ollama" class="mention hashtag" rel="nofollow noopener" target="_blank">#ollama</a> <a href="https://techhub.social/tags/embedding" class="mention hashtag" rel="nofollow noopener" target="_blank">#embedding</a> <a href="https://techhub.social/tags/reranking" class="mention hashtag" rel="nofollow noopener" target="_blank">#reranking</a> <a href="https://techhub.social/tags/golang" class="mention hashtag" rel="nofollow noopener" target="_blank">#golang</a> <a href="https://techhub.social/tags/ai" class="mention hashtag" rel="nofollow noopener" target="_blank">#ai</a> <a href="https://techhub.social/tags/llm" class="mention hashtag" rel="nofollow noopener" target="_blank">#llm</a>

Rost GlukhovQwen3 Embedding & Reranker Models on Ollama: State-of-the-Art Performance <a href="https://www.glukhov.org/post/2025/06/qwen3-embedding-qwen3-reranker-on-ollama/" rel="nofollow noopener" translate="no" target="_blank">https://www.glukhov.org/post/2025/06/qwen3-embedding-qwen3-reranker-on-ollama/</a> <a href="https://techhub.social/tags/Qwen3" class="mention hashtag" rel="nofollow noopener" target="_blank">#Qwen3</a> <a href="https://techhub.social/tags/Embedding" class="mention hashtag" rel="nofollow noopener" target="_blank">#Embedding</a> <a href="https://techhub.social/tags/Reranker" class="mention hashtag" rel="nofollow noopener" target="_blank">#Reranker</a> <a href="https://techhub.social/tags/LLM" class="mention hashtag" rel="nofollow noopener" target="_blank">#LLM</a> <a href="https://techhub.social/tags/AI" class="mention hashtag" rel="nofollow noopener" target="_blank">#AI</a> <a href="https://techhub.social/tags/ollama" class="mention hashtag" rel="nofollow noopener" target="_blank">#ollama</a>

Inautilo<a href="https://mastodon.social/tags/Development" class="mention hashtag" rel="nofollow noopener" target="_blank">#Development</a> <a href="https://mastodon.social/tags/Techniques" class="mention hashtag" rel="nofollow noopener" target="_blank">#Techniques</a> Introducing php-node · How to seamlessly blend PHP with Node.js <a href="https://ilo.im/164g4x" rel="nofollow noopener" translate="no" target="_blank">https://ilo.im/164g4x</a>_____ <a href="https://mastodon.social/tags/Programming" class="mention hashtag" rel="nofollow noopener" target="_blank">#Programming</a> <a href="https://mastodon.social/tags/Coding" class="mention hashtag" rel="nofollow noopener" target="_blank">#Coding</a> <a href="https://mastodon.social/tags/Embedding" class="mention hashtag" rel="nofollow noopener" target="_blank">#Embedding</a> <a href="https://mastodon.social/tags/NodeJS" class="mention hashtag" rel="nofollow noopener" target="_blank">#NodeJS</a> <a href="https://mastodon.social/tags/PHP" class="mention hashtag" rel="nofollow noopener" target="_blank">#PHP</a> <a href="https://mastodon.social/tags/WordPress" class="mention hashtag" rel="nofollow noopener" target="_blank">#WordPress</a> <a href="https://mastodon.social/tags/CMS" class="mention hashtag" rel="nofollow noopener" target="_blank">#CMS</a> <a href="https://mastodon.social/tags/WebDev" class="mention hashtag" rel="nofollow noopener" target="_blank">#WebDev</a> <a href="https://mastodon.social/tags/Frontend" class="mention hashtag" rel="nofollow noopener" target="_blank">#Frontend</a> <a href="https://mastodon.social/tags/Backend" class="mention hashtag" rel="nofollow noopener" target="_blank">#Backend</a>

Gea-Suan Lin<a href="https://blog.gslin.org/archives/2025/06/09/12444/mariadb-11-8-lts/" rel="nofollow noopener" translate="no" target="_blank">https://blog.gslin.org/archives/2025/06/09/12444/mariadb-11-8-lts/</a>MariaDB 11.8 LTS#2038 #2106 <a href="https://abpe.org/tags/database" class="mention hashtag" rel="nofollow noopener" target="_blank">#database</a> <a href="https://abpe.org/tags/embedding" class="mention hashtag" rel="nofollow noopener" target="_blank">#embedding</a> <a href="https://abpe.org/tags/hnsw" class="mention hashtag" rel="nofollow noopener" target="_blank">#hnsw</a> <a href="https://abpe.org/tags/mariadb" class="mention hashtag" rel="nofollow noopener" target="_blank">#mariadb</a> <a href="https://abpe.org/tags/mysql" class="mention hashtag" rel="nofollow noopener" target="_blank">#mysql</a> <a href="https://abpe.org/tags/problem" class="mention hashtag" rel="nofollow noopener" target="_blank">#problem</a> <a href="https://abpe.org/tags/rdbms" class="mention hashtag" rel="nofollow noopener" target="_blank">#rdbms</a> <a href="https://abpe.org/tags/signed" class="mention hashtag" rel="nofollow noopener" target="_blank">#signed</a> <a href="https://abpe.org/tags/timestamp" class="mention hashtag" rel="nofollow noopener" target="_blank">#timestamp</a> <a href="https://abpe.org/tags/unsigned" class="mention hashtag" rel="nofollow noopener" target="_blank">#unsigned</a> <a href="https://abpe.org/tags/vector" class="mention hashtag" rel="nofollow noopener" target="_blank">#vector</a> <a href="https://abpe.org/tags/year" class="mention hashtag" rel="nofollow noopener" target="_blank">#year</a>

HabrMem-векторы: как сохранить 1500 токенов в одном векторе и зачем это нужноКаждый, кто работал с большими языковыми моделями (LLM), знает про ограничение длины контекста: модель не может напрямую обработать текст, превышающий определённое число токенов. Это накладывает ограничения на работу с длинными документами и обширным контекстом. Но что если бы мы могли упаковать длинный текст в один-единственный вектор и скормить его модели как обычный токен? Звучит фантастично, однако свежие исследования показывают, что это возможно – такие “mem-векторы” позволяют сохранить сотни и даже полторы тысячи токенов информации в одном эмбеддинге. Это принципиально иной подход, нежели классическое сжатие данных, и он сулит интересные применения. Mem-вектор (от “memory vector”) – это специально обученный вектор, который хранит содержание целого текста. Идея в том, что если модель умеет предсказывать текст, то можно подобрать такой вектор на входе, при котором замороженная (неизменяемая) LLM сама декодирует исходный текст . Иначе говоря, mem-вектор играет роль «семени», из которого предобученная модель порождает заложенное в нём сообщение. В этой статье разберём, как это работает, почему вообще возможно “запихнуть” роман в один вектор и какие ограничения при этом появляются. Также сравним mem-подход с классическими алгоритмами сжатия (Huffman, арифметическое кодирование, zlib и др.), обсудим последние научные работы на эту тему и возможные применения: от Retrieval-Augmented Generation (RAG) до передачи новых знаний замороженным моделям. Центральная мысль: mem-векторы – это не просто компрессия текста, а способ напрямую скормить модели смысл и знания, минуя последовательное чтение токенов . Разбираемся далее<a href="https://habr.com/ru/articles/906592/" rel="nofollow noopener" translate="no" target="_blank">https://habr.com/ru/articles/906592/</a><a href="https://zhub.link/tags/mem%D0%B2%D0%B5%D0%BA%D1%82%D0%BE%D1%80" class="mention hashtag" rel="nofollow noopener" target="_blank">#memвектор</a> <a href="https://zhub.link/tags/llm" class="mention hashtag" rel="nofollow noopener" target="_blank">#llm</a> <a href="https://zhub.link/tags/%D1%8D%D0%BD%D1%82%D1%80%D0%BE%D0%BF%D0%B8%D1%8F_%D1%82%D0%B5%D0%BA%D1%81%D1%82%D0%B0" class="mention hashtag" rel="nofollow noopener" target="_blank">#энтропия_текста</a> <a href="https://zhub.link/tags/%D0%BA%D0%BE%D0%BC%D0%BF%D1%80%D0%B5%D1%81%D1%81%D0%B8%D1%8F" class="mention hashtag" rel="nofollow noopener" target="_blank">#компрессия</a> <a href="https://zhub.link/tags/RAG" class="mention hashtag" rel="nofollow noopener" target="_blank">#RAG</a> <a href="https://zhub.link/tags/embedding" class="mention hashtag" rel="nofollow noopener" target="_blank">#embedding</a> <a href="https://zhub.link/tags/hidden_capacity" class="mention hashtag" rel="nofollow noopener" target="_blank">#hidden_capacity</a> <a href="https://zhub.link/tags/NLP" class="mention hashtag" rel="nofollow noopener" target="_blank">#NLP</a>

JMLR'Variance-Aware Estimation of Kernel Mean Embedding', by Geoffrey Wolfer, Pierre Alquier.<a href="http://jmlr.org/papers/v26/23-0161.html" rel="nofollow noopener" translate="no" target="_blank">http://jmlr.org/papers/v26/23-0161.html</a> <a href="https://sigmoid.social/tags/embeddings" class="mention hashtag" rel="nofollow noopener" target="_blank">#embeddings</a> <a href="https://sigmoid.social/tags/embedding" class="mention hashtag" rel="nofollow noopener" target="_blank">#embedding</a> <a href="https://sigmoid.social/tags/empirical" class="mention hashtag" rel="nofollow noopener" target="_blank">#empirical</a>

Unofficial PetaPixel BotPhotographer Asks Supreme Court to Decide if Embedded Instagram Posts Infringe Copyright <a href="https://petapixel.com/2025/04/09/photographer-asks-supreme-court-to-decide-if-embedded-instagram-posts-infringe-copyright/" rel="nofollow noopener" translate="no" target="_blank">https://petapixel.com/2025/04/09/photographer-asks-supreme-court-to-decide-if-embedded-instagram-posts-infringe-copyright/</a> <a href="https://toot.earth/tags/copyrightinfringement" class="mention hashtag" rel="nofollow noopener" target="_blank">#copyrightinfringement</a> <a href="https://toot.earth/tags/embedfeature" class="mention hashtag" rel="nofollow noopener" target="_blank">#embedfeature</a> <a href="https://toot.earth/tags/supremecourt" class="mention hashtag" rel="nofollow noopener" target="_blank">#supremecourt</a> <a href="https://toot.earth/tags/Technology" class="mention hashtag" rel="nofollow noopener" target="_blank">#Technology</a> <a href="https://toot.earth/tags/embedding" class="mention hashtag" rel="nofollow noopener" target="_blank">#embedding</a> <a href="https://toot.earth/tags/instagram" class="mention hashtag" rel="nofollow noopener" target="_blank">#instagram</a> <a href="https://toot.earth/tags/lawsuit" class="mention hashtag" rel="nofollow noopener" target="_blank">#lawsuit</a> <a href="https://toot.earth/tags/News" class="mention hashtag" rel="nofollow noopener" target="_blank">#News</a> <a href="https://toot.earth/tags/Law" class="mention hashtag" rel="nofollow noopener" target="_blank">#Law</a>

Toni AittoniemiHow to speak in ways AI bots won’t understand.今朝毎朝ボット<a href="https://youtube.com/watch?v=F4KQ8wBt1Qg" rel="nofollow noopener" translate="no" target="_blank">https://youtube.com/watch?v=F4KQ8wBt1Qg</a> <a href="https://mastodon.green/tags/ai" class="mention hashtag" rel="nofollow noopener" target="_blank">#ai</a> <a href="https://mastodon.green/tags/llm" class="mention hashtag" rel="nofollow noopener" target="_blank">#llm</a> <a href="https://mastodon.green/tags/machinelearning" class="mention hashtag" rel="nofollow noopener" target="_blank">#machinelearning</a> <a href="https://mastodon.green/tags/embedding" class="mention hashtag" rel="nofollow noopener" target="_blank">#embedding</a> <a href="https://mastodon.green/tags/context" class="mention hashtag" rel="nofollow noopener" target="_blank">#context</a>

Hacker NewsSOTA Code Retrieval with Efficient Code Embedding Models — <a href="https://www.qodo.ai/blog/qodo-embed-1-code-embedding-code-retreival/" rel="nofollow noopener" translate="no" target="_blank">https://www.qodo.ai/blog/qodo-embed-1-code-embedding-code-retreival/</a> <a href="https://mastodon.social/tags/HackerNews" class="mention hashtag" rel="nofollow noopener" target="_blank">#HackerNews</a> <a href="https://mastodon.social/tags/SOTA" class="mention hashtag" rel="nofollow noopener" target="_blank">#SOTA</a> <a href="https://mastodon.social/tags/Code" class="mention hashtag" rel="nofollow noopener" target="_blank">#Code</a> <a href="https://mastodon.social/tags/Retrieval" class="mention hashtag" rel="nofollow noopener" target="_blank">#Retrieval</a> <a href="https://mastodon.social/tags/Code" class="mention hashtag" rel="nofollow noopener" target="_blank">#Code</a> <a href="https://mastodon.social/tags/Embedding" class="mention hashtag" rel="nofollow noopener" target="_blank">#Embedding</a> <a href="https://mastodon.social/tags/AI" class="mention hashtag" rel="nofollow noopener" target="_blank">#AI</a> <a href="https://mastodon.social/tags/Technology" class="mention hashtag" rel="nofollow noopener" target="_blank">#Technology</a> <a href="https://mastodon.social/tags/Machine" class="mention hashtag" rel="nofollow noopener" target="_blank">#Machine</a> <a href="https://mastodon.social/tags/Learning" class="mention hashtag" rel="nofollow noopener" target="_blank">#Learning</a>

Alejandro Duarte<a href="https://mastodon.online/tags/AI" class="mention hashtag" rel="nofollow noopener" target="_blank">#AI</a> and <a href="https://mastodon.online/tags/RAG" class="mention hashtag" rel="nofollow noopener" target="_blank">#RAG</a> - Learning the basics: What exactly is an <a href="https://mastodon.online/tags/embedding" class="mention hashtag" rel="nofollow noopener" target="_blank">#embedding</a> and how to use them in <a href="https://mastodon.online/tags/MariaDB" class="mention hashtag" rel="nofollow noopener" target="_blank">#MariaDB</a>? <a href="https://www.youtube.com/watch?v=XkB2DLK60JU" rel="nofollow noopener" translate="no" target="_blank">https://www.youtube.com/watch?v=XkB2DLK60JU</a>

linkdropGitHub - lancedb/lancedb: Developer-friendly, serverless vector database for AI applications. Easily add long-term memory to your LLM apps! <a href="https://github.com/lancedb/lancedb" rel="nofollow noopener" translate="no" target="_blank">https://github.com/lancedb/lancedb</a> <a href="https://troet.cafe/tags/persistence" class="mention hashtag" rel="nofollow noopener" target="_blank">#persistence</a> <a href="https://troet.cafe/tags/OpenSource" class="mention hashtag" rel="nofollow noopener" target="_blank">#OpenSource</a> <a href="https://troet.cafe/tags/embedding" class="mention hashtag" rel="nofollow noopener" target="_blank">#embedding</a> <a href="https://troet.cafe/tags/database" class="mention hashtag" rel="nofollow noopener" target="_blank">#database</a> <a href="https://troet.cafe/tags/GitHub" class="mention hashtag" rel="nofollow noopener" target="_blank">#GitHub</a> <a href="https://troet.cafe/tags/search" class="mention hashtag" rel="nofollow noopener" target="_blank">#search</a> <a href="https://troet.cafe/tags/vector" class="mention hashtag" rel="nofollow noopener" target="_blank">#vector</a> <a href="https://troet.cafe/tags/ai" class="mention hashtag" rel="nofollow noopener" target="_blank">#ai</a>

onionSo basically face recognision is: compare current <a href="https://mastodon.tal.org/tags/embedding" class="mention hashtag" rel="nofollow noopener" target="_blank">#embedding</a> distance to database of embeddings and closest is considered a match? <a href="https://mastodon.tal.org/tags/facerecognition" class="mention hashtag" rel="nofollow noopener" target="_blank">#facerecognition</a> <a href="https://mastodon.tal.org/tags/opencv" class="mention hashtag" rel="nofollow noopener" target="_blank">#opencv</a>

Piotr MigdałI’m excited to share my newest blog post, "Don't sure cosine similarity carelessly"<a href="https://p.migdal.pl/blog/2025/01/dont-use-cosine-similarity" rel="nofollow noopener" translate="no" target="_blank">https://p.migdal.pl/blog/2025/01/dont-use-cosine-similarity</a>We often rely on cosine similarity to compare embeddings—it's like “duct tape” for vector comparisons. But just like duct tape, it can quietly mask deeper problems. Sometimes, embeddings pick up a “wrong kind” of similarity, matching questions to questions instead of questions to answers or getting thrown off by formatting quirks and typos rather than the text's real meaning.In my post, I discuss what can go wrong with off-the-shelf cosine similarity and share practical alternatives. If you’ve ever wondered why your retrieval system returns oddly matched items or how to refine your embeddings for more meaningful results, this is for you! ` I want to thank Max Salamonowicz and Grzegorz Kossakowski for their feedback after my flash talk at the Warsaw AI Breakfast, Rafał Małanij for inviting me to give a talk at the Python Summit, and for all the curious questions at the conference, and LinkedIn. <a href="https://mathstodon.xyz/tags/cosineSimilarity" class="mention hashtag" rel="nofollow noopener" target="_blank">#cosineSimilarity</a> <a href="https://mathstodon.xyz/tags/embedding" class="mention hashtag" rel="nofollow noopener" target="_blank">#embedding</a> <a href="https://mathstodon.xyz/tags/llm" class="mention hashtag" rel="nofollow noopener" target="_blank">#llm</a> <a href="https://mathstodon.xyz/tags/similarity" class="mention hashtag" rel="nofollow noopener" target="_blank">#similarity</a>

RicardoDamn, this is really cool, but I wish it had a big “pre-requisites” in the readme with “NVIDIA” in it <a href="https://mstdn.social/tags/AI" class="mention hashtag" rel="nofollow noopener" target="_blank">#AI</a> <a href="https://mstdn.social/tags/RAG" class="mention hashtag" rel="nofollow noopener" target="_blank">#RAG</a> <a href="https://mstdn.social/tags/Embedding" class="mention hashtag" rel="nofollow noopener" target="_blank">#Embedding</a> <a href="https://mstdn.social/tags/Documents" class="mention hashtag" rel="nofollow noopener" target="_blank">#Documents</a> <a href="https://mstdn.social/tags/Ollama" class="mention hashtag" rel="nofollow noopener" target="_blank">#Ollama</a> <a href="https://github.com/TilmanGriesel/chipper" rel="nofollow noopener" translate="no" target="_blank">https://github.com/TilmanGriesel/chipper</a>

Recent searches

Search options

Administered by:

Server stats:

#embedding