₿abakar LangChain If you are interested in end to end vector search without worrying about managing embeddings, then check out Marqo . Documents in, documents out. github.com/marqo-ai/marqo…

It is absolutely insane and idiotic that most people are rolling their own embedding and vector storage instead of just using marqo.ai on their AI projects. Oh yeah, most of them are using Python. Guess it makes sense then... 😂 #gpt #ai #langchain #openai

Our new demo is live - to celebrate, we're giving away some FREE MERCH 🎉 To win a Marqo hoodie: 1. Check out the demo (demo.marqo.ai) 2. Screenshot a weird piece of AI-generated fashion that you come across 3. Reply to this tweet with the screenshot 4. Retweet this!

I've learnt so much from jesse - pumped to have him on the Wild Hearts pod sharing the magic of Marqo 🦛 open.spotify.com/episode/5bsiDV…

Computer science programmes at university are at the forefront of much of the innovation in LLMs. Great to host two of our AI founders jesse and Iain Mackie for an Large Language evening at CF.

Welcome to the Haystack family Marqo 🙌 Thank you for the contribution @owen_p_elliott, it's great to see the support for more technologies growing for our community 🚀 The fresh-out-of-the-oven MarqoDocumentStore is available for Haystack 2.0 which you can already start

This week is packed with news, and we have just started. Marqo announces Marqo V2 using vespa.ai! Read on why they chose Vespa. marqo.ai/blog/marqo-v2-…

📈CLIP benchmarks for multi-modal search and recommendations 📊Updated benchmarks for over one hundred CLIP models to assess them not only on text to image retrieval (a proxy for search relevancy) but also: - Latency when embedding text or images. - Text context length. -

Great to see Marqo (and open source) featured prominently in the latest TechCrunch article on the role and rise of vector databases in AI. Although the hype, getting vector search into production is still very challenging. Vector search requires much more than vector

We are excited to announce that Marqo is now avaliable on Google Cloud Marketplace! 🌐✨ Google Cloud Google Cloud ANZ Why Choose Marqo on Google Cloud? ✔ Scalable Architecture ✔ High Availability ✔ Free license More about this: marqo.ai/blog/vector-se…

Matryoshka representation learning (MRL) for CLIP (&SigLip) Key findings: - Reducing embeddings size by 4x retains ~95 performance across in and out of domain on multi-modal retrieval - Works with rank-tuning methods like GCL - MRL trained models can improve if the original

The Impacts of Data, Ordering, and Intrinsic Dimensionality on Recall in Hierarchical Navigable Small Worlds Explores HNSW algorithm on real-world vector embeddings, revealing a significant impact of intrinsic dimensionality and data insertion order. 📝arxiv.org/abs/2405.17813

🎉 Announcing Marqo 2.7.0! 🎉 New features: 🌟 Updated Open CLIP version and support of new families of models 🌟 New and state-of-the-art multimodal models. You can choose these models to build your index! 🌟 Support of lexical search with only a filter. Marqo now supports

I’ve had an incredible time in San Francisco for the past 3 weeks, meeting the Marqo team and kickstarting my new role in the #AI & #Tech industry!🚀 Here are some of the highlights in thread form 🧵

HNSW recall can severely degrade vector search results due to the interactions between the dataset topology and ordering, embedding models, and approximate KNN algorithms 🧵 + Defaults matter: implementations from popular vector databases degrade NDCG by -18%. + Order matters:

Join our Co-Founder and CTO, jesse, at Relevance AI this Wednesday! He's part of a great line-up for January Capital x @the_builderclub's panel on 'The Reality of Fundraising'!

As part of our Ecommerce Embedding Model launch, we also released the following datasets: 📊 Marqo/amazon-products-eval 📊 Marqo/google-shopping-general-eval 📊 Marqo/amazon-products-eval-100k 📊 Marqo/google-shopping-general-eval-100k More in the Hugging Face collection:

🚀 New Marqo Ecommerce Models Outperform Amazon by up to 88%! 🚀 We're thrilled to release two new foundation models for multimodal product embeddings, Marqo-Ecommerce-B and Marqo-Ecommerce-L! Learn more: marqo.ai/blog/introduci…

Today Marqo open-weight (Apache 2.0) released the two best embedding models for ecommerce search and recommendations available anywhere. Marqo ecommerce models significantly outperform models from Amazon, Google, Cohere and Jina (see below). Fun fact: we had to create a

My observation is that GEO has so far delivered little to no measurable benefit for ecommerce. Consumers still overwhelmingly prefer browsing a digital storefront—where brands can control the experience, own the data, and nurture the relationship.