{"slug": "fanout-vectors", "title": "Fanout Vectors", "summary": "SEO practitioner Dejan published a reproduction of Google's query fanout research, training a vector-based diffusion model that replaces text fanouts with embeddings for document retrieval and releasing it on Hugging Face at dejanseo/fanout-diffusion with a companion demo Space. The reproduction skipped reinforcement learning, using Gemma-4-26B-A4B as a teacher model to generate synthetic orthogonal fanout training data for Google's ecommerce taxonomy, and achieved what the author describes as blazing speeds via a 1-bit model architecture and custom CUDA kernels. The author argues Google could run this approach far more cheaply and efficiently in production, where text fanouts would be eliminated.", "body_md": "An analysis and reproduction of Google's query fanout research, demonstrating how vector-based diffusion models could replace text fanouts for faster search.\n\nLast week [I wrote about Google's new query fanout research piece](https://dejan.ai/blog/fanout2/) and [made a prediction that if this goes to production text fanouts are gone](https://lnkd.in/p/g9wCDWNB).\n\nTwo things:\n\nI reproduced the paper and trained the model myself. I also decided to simplify the process and skip RL which I didn't see as necessary step considering we now have excellent teacher models such as Gemma-4-26B-A4B suitable for synthetic training data generation for quality orthogonal fanouts.\n\nThe trained model is available here:[https://huggingface.co/dejanseo/fanout-diffusion](https://huggingface.co/dejanseo/fanout-diffusion)\n\nAnd the demo is here:[https://huggingface.co/spaces/dejanseo/fanout-diffusion](https://huggingface.co/spaces/dejanseo/fanout-diffusion)\n\nThe above demo employs the trained diffusion vector fanout model and instead of text queries uses embeddings for document retrieval.\n\nThe dataset generated is for Google's ecommerce taxonomy fanouts. Obviously for a production model you'd go beyond e-commerce and train on a much larger data sample.\n\nWhat's interesting though is the blazing speeds I achieved with 1-bit model architecture and custom CUDA kernels. I think Google could realistically achieve that level of wizardry on their and and make things run much cheaper and more efficiently.", "url": "https://wpnews.pro/news/fanout-vectors", "canonical_source": "https://dejan.ai/blog/fanout-vectors/", "published_at": "2026-09-24 14:24:05+00:00", "updated_at": "2026-09-24 14:32:23.169993+00:00", "lang": "en", "topics": ["ai-search", "generative-engine-optimization", "artificial-intelligence", "machine-learning", "generative-ai"], "entities": ["Google", "Dejan", "Gemma-4-26B-A4B", "Hugging Face", "dejanseo/fanout-diffusion"], "also_reported_by": [], "alternates": {"html": "https://wpnews.pro/news/fanout-vectors", "markdown": "https://wpnews.pro/news/fanout-vectors.md", "text": "https://wpnews.pro/news/fanout-vectors.txt", "jsonld": "https://wpnews.pro/news/fanout-vectors.jsonld"}}