
https://teapaw.org/t.co/XEw3GZ9G7m CTO (@vespaengine) and long-time (now) ex-Yahoo. Nerd. Father. And more...
Trondheim, Norway
Joined July 2007
- Tweets3.2K
- Following322
- Followers258
- Likes301
Frode Lundgren retweeted
Let's make search great again 🚀
I'm joining Vespa ai and my role is to help you build a world class search, RAG and recommendation systems.
In addition, I will be:
👨🏽💻 Taking part in open-source communities and contributions
👀 Creating tutorials to learn how to build search for real world problems
📈 Working with cross-functional teams to bring search from "good" to "great" for real users
👨🏽🎓 Learning and evolving: as always
If you’re passionate about search, rag, embeddings, hybrid architectures, or you’re using (or exploring) Vespa ai and want to collaborate or chat, please reach out! Always happy to connect.
Frode Lundgren retweeted
Perplexity builds AI Search on Vespa.ai
Blog post below 👇
Frode Lundgren retweeted
A new blog post on how stress-free your life is when you run your search on @vespaengine even at a billion-scale!
Frode Lundgren retweeted
2024 in review:
- The rise of VLMs for retrieval (ColPali)
- The year of late-interaction (Col)
- (V)LMs as rankers
- Token cost race to zero + context cache
- RAG is dead, long live RAG
- Vector db as a category died
- The rise of BM25
- BQ Binarization rediscovered
Frode Lundgren retweeted
🚀 The Rise of Vision RAG!
Launching a complete RAG app that you can deploy to production in minutes!
- Hybrid fusion of ColPali + BM25 with @vespaengine
- Gemini 1.5 Flash-8B
- FastHTML frontend
- Runs on Huggingface Spaces
Interpretable SERP with snippets + patch highlights!
RAG with ColPali doesn't need to be sluggish.
Huge s/o to the team that built it @thomas_thoresen @andreer @ldalves
Frode Lundgren retweeted
Matryoshka 🤝 Binary vectors: Slash vector search costs with Vespa
We announce support for combining matryoshka and binary quantization in Vespa’s native hugging-face embedder and discuss how this slashes vector search costs.
blog.vespa.ai/combining-matr…
Frode Lundgren retweeted
Seems everybody is migrating their search and recommendation systems from Elastic to Vespa now.
Here's the experience of Stanby, Japan's leading job search site: blog.vespa.ai/migrating-to-t…
Frode Lundgren retweeted
Cohere Embed V3 Binary Vectors + Vespa
We got excited over the new Cohere embedding models, which slash your vector DB cost 4x—32x. SOTA models with just 128 bytes per vector.
Here are three comprehensive guides on using them with @vespaengine:
blog.vespa.ai/scaling-large-…
Frode Lundgren retweeted
This week is packed with news, and we have just started. @marqo_ai announces Marqo V2 using @vespaengine! Read on why they chose Vespa.
marqo.ai/blog/marqo-v2-perfo…
Frode Lundgren retweeted
Binary embeddings from @cohere with @vespaengine!
- HNSW index with hamming distance over 1024 bits!
- Re-ranking with the dot product between full query vector (1024 floats) against an unpacked float version of the binary embedding.
Notebook:
pyvespa.readthedocs.io/en/la…
Frode Lundgren retweeted
The Singaporean government has deployed Vespa to search every word ever said in their Parliament.
"A good decision is an informed one [...] The heart of a good RAG system is a good search engine to retrieve the relevant data chunks for ingestion"
Many teams are racing to make use of these new methods to achieve superior quality, but the Singaporean government may have been first to put them in production!
Frode Lundgren retweeted
I believe that M3 is a welcome direction in training embedding models.
A single forward pass through the model obtains several representations. Highest possible flexibility to express the model in retrieval and ranking pipelines!
Get started with M3 colab.research.google.com/gi…
Frode Lundgren retweeted
The Mother of all Embedding Models just dropped!
Embeds text into three optional representations.
- ColBERT (multi-vector, one per token)
- sparse (splade-like)
- dense (your regular single vector embedding)
We got you covered with @vespaengine
BGE-M3 🔥 huggingface.co/BAAI/bge-m3
Highlights:
✨ Multi-Lingual: supports over 100 languages
✨ Multi-Granularity: input texts up to 8192 characters
✨ Multi-Functionality: integrates dense retrieval/ sparse retrieval/ multi-vector retrieval to support different scenarios
Frode Lundgren retweeted
⚡️Turbocharge RAG with LangChain and Vespa Streaming Mode for Sharded Data
@vespaengine's streaming search solution enables you to efficiently search over sharded data - perfect for when you want to store data grouped by user id
@jobergum wrote a great article (complete with code) on how to use this in a custom LangChain chain
This involves a custom retriever and a custom chain constructed with LangChain Expression Language
Blog: blog.vespa.ai/turbocharge-ra…
Code: colab.research.google.com/gi…
Frode Lundgren retweeted
Follow this guide if you need to build a personal data app which
- costs about 5% of a traditional vector database
- contrary to those, finds all the user's data
- scales to millions (or billions) of users
Hands-on RAG guide for personal data with @vespaengine and @llama_index 😍
Featuring LLamaIndex retrievers, Vespa streaming mode for personal data, built-in embedders, multi-index RAG (federation + blend), hybrid search, and rank fusion strategies.
blog.vespa.ai/scaling-person…
Frode Lundgren retweeted
Today, we announce a new way to search Vespa content - powered by Vespa.
This blog post shares why, how, and what we learned.
blog.vespa.ai/search-vespa-a…
Frode Lundgren retweeted
Semantic search with multiple languages is finally clearing the bar for production.
Vespa makes it easy to stand up your own.
blog.vespa.ai/simplify-searc…
For those who have been around the search space for some time, "personal search" used to be all the rage until it weren't that many usecases that couldn't be done with 'grep'. With embeddings, LLMs, and generative AI, a new need has emerged, and Vespa has you covered. Of course.
Vector databases are not suited for personal data: You need complete results and much lower cost.
We're announcing vector streaming search to solve this problem: blog.vespa.ai/announcing-vec…
Fully production ready, and open source, of course.


















