Category: Google

  • Top 10 Most Recent Papers by MUVERA Authors

    Top 10 Most Recent Papers by MUVERA Authors

    —

    by

    Dan Petrovic
    in , ,

    MUVERA Authors: 1. Laxman Dhulipala (Google Research & UMD) Top 10 Recent Papers (2023-2025) Research Focus Areas 2. Majid Hadian (Google DeepMind) Top 10 Recent Papers (2023-2025) Research Focus Areas 3. Jason Lee (Google Research & UC Berkeley) Top 10 Recent Papers (2023-2025) Research Focus Areas 4. Rajesh Jayaram (Google Research) Top 10 Recent Papers…

  • Training Gemma‑3‑1B Embedding Model with LoRA

    Training Gemma‑3‑1B Embedding Model with LoRA

    —

    by

    Dan Petrovic
    in , ,

    In our previous post, Training a Query Fan-Out Model, we demonstrated how to generate millions of high-quality query reformulations without human labelling, by navigating the embedding space between a seed query and its target document and then decoding each intermediate vector back into text using a trained query decoder. That decoder’s success critically depends on…

  • Training a Query Fan-Out Model

    Training a Query Fan-Out Model

    —

    by

    Dan Petrovic
    in , ,

    Google discovered how to generate millions of high-quality query reformulations without human input by literally traversing the mathematical space between queries and their target documents. Here’s How it Works This generated 863,307 training examples for a query suggestion model (qsT5) that outperforms all existing baselines. Query Decoder + Latent Space Traversal Step 1: Build a…

  • Cosine Similarity or Dot Product?

    Cosine Similarity or Dot Product?

    —

    by

    Dan Petrovic
    in ,

    Google’s embedder uses dot product between normalized vectors which is computationally more efficient but mathematically equivalent to cosine similarity. How Googler’s work and think internally typically aligns with their open source code (Gemini -> Gemma) and Chrome is no exception. It’s why I look there for answers and clarity on Google’s machine learning approaches. After…

  • Dissecting Gemini’s Tokenizer and Token Scores

    Dissecting Gemini’s Tokenizer and Token Scores

    —

    by

    Dan Petrovic
    in ,

    As a technical SEO, you might be diving into machine learning (ML) to understand how tools like Google’s Gemini process text. One foundational concept is subword tokenization—breaking words into smaller pieces called “tokens.” While tokens themselves are context-agnostic (they don’t consider surrounding words), they do carry an inherent bias: each token’s likelihood reflects how prominent…

  • There’s a small army of on-device models coming to Chrome

    There’s a small army of on-device models coming to Chrome

    —

    by

    Dan Petrovic
    in ,

    1. ULM128M 2. LLMIT1B 3. GEMINI2_NANOV2 4. GEMINI2_NANOV2_EE2Q 5. GEMINI_XS 6. GEMINI_XS_DRAFTER_6LAYER_CAUSAL_USM_700M_RESIDUAL 7. GEMINI_XS_LUSM_700M_RESIDUAL_BOTTOM15 8. GEMINI2_NANOV2_EE12Q 9. GEMINI2_NANOV2_EE2_LUSM_700M 10. GEMINI2_NANOV2_CAUSAL_700M 11. GEMINI2_NANOV2_EE20_CAUSAL_LUSM_700M 12. GEMINI_XL_DRAFTER_24LAYER 13. GEMINI_XS_FA1 14. GEMMA2_8B 15. GEMMA2_7B 16. GEMMA2_2B 17. GEMMA3_1B 18. GEMMA3_4B 19. GEMMA3_12B 20. GEMMA3_27B 21. STABLELM_4E1T_3B_PHI_2_TF_LITE

  • Query Fan-Out Prompt Implementation in Google’s Open-Source Agentic Framework

    Query Fan-Out Prompt Implementation in Google’s Open-Source Agentic Framework

    —

    by

    Dan Petrovic
    in ,

    Google’s open-source “Gemini Fullstack LangGraph Quickstart” pairs Gemini 2.5 with LangGraph to showcase a fully transparent, citation-driven research agent (Mikami 2025). A React frontend (Vite, Tailwind CSS, Shadcn UI) collects user queries and displays progress, while a FastAPI/LangGraph backend orchestrates a multi-step workflow: Although this isn’t Google’s official Gemini implementation as seen in AI Mode…

  • AI Mode & Page Indexing

    AI Mode & Page Indexing

    —

    by

    Dan Petrovic
    in ,

    Our tests show that Google’s AI Mode doesn’t retrieve page content from the live web but somewhere else, and that “somewhere else” appears to be a proprietary content store separate from the search index. How do we know this? We found a case where AI Mode failed to fetch a page that’s indexed and ranking…

  • AI Mode is Not Live Web

    AI Mode is Not Live Web

    —

    by

    Dan Petrovic
    in ,

    I recently stumbled upon a fascinating aspect of how Google’s AI Mode (powered by a custom Gemini model) interacts with the internet. I ran a simple test, and the results suggest that instead of performing truly live fetches for all URLs, the AI Mode relies on Google’s existing index or a cached version of the…

  • How AI Mode Selects Snippets

    How AI Mode Selects Snippets

    —

    by

    Dan Petrovic
    in ,

    I noticed out commented out bits in the source code of the AI Mode results. They contain actual snippets supplied to Gemini to form the response. This is not what is displayed to the user. It’s what search tool supplies to Gemini which then renders the response to the user. This is kind of a…