🔥The 9th Round of Easy Loan, Earn $40 Reward is in progress❗️ ⏰ Promotion Period: January 15th - Feburary 15th, 2025 👉 Register now and check more details at gate.io/campaigns/358

The Weaviate Recommender Service is going to Beta testing! 🎉🎉🎉 It has been nearly 2 years (!!) since we launched our initial Recommendation focused Weaviate feature, Ref2Vec. Ref2Vec represents a user as the average vector of the product vectors they have interacted with. 🥗

Nature is Amazing ☘️ Duck runs in a marathon and gets a medal

The cross-encoder is a popular architecture for re-ranking models. Here’s how they work: 1. Concatenate query and document ([CLS] Query [SEP] Document [SEP]) 2. Feed through transformer layers 3. Use [CLS] token representation to score relevance Turn good results into better

1 + 1 = 14 The whole is always more than the sum of its parts. As Weaviate grew, I evolved too—taking on roles in branding, marketing, design, growth management, and more. We’ve gone from a startup to a force of nature. Two years ago, I joined Weaviate, and it’s been a crazy

25x faster filtered search queries in 1.27? 🤯 With Weaviate’s new ACORN filtered search strategy in v1.27, we’re seeing filtered search queries that are way faster than they were before! It improves upon the existing strategy to better handle negatively correlated queries and

there was a brain grenade lofted my way today... "what's the coolest basketball data visualization you've seen someone else create?" 💥🫨 can't pick one, so here's a thread of a few of my all-time favs, starting with these 3D-printed shot charts from Nathan Yau 😍😍😍

Have you wondered how keyword search works at Weaviate? This blog post shows how it works from the inside out and the recent 1.29 improvements with BlockMax WAND that led to 90%+ performance gains. Check out the blog post and demo by André Mourão and JP Hwang:

One vector is not enough to capture meaning. Multi-vector embeddings are changing vector search forever. Here’s how: If you're familiar with vector embeddings, you know how they're used to transform data (like text or images) into a numerical format that machine learning models

You probably know vector embeddings. But do you know the difference between single and multi-vector embeddings? 🤔 Single-vector embeddings are the traditional type of embeddings. They generate one embedding for an entire input (like a document) to represent it in tasks such as

Dallas, we love you! 🤠 Thank you to everyone who came out to AI [in Prod]! We had a packed day of tech talks from our incredible experts including Sandeep Kulkarni from AWS and Michael Pont from LoyeeAI. 💚 We might work in AI, but it takes real humans to put these events

Ayush S When I was first getting started my go to was a8e08566-b341-4e4c-9f44-8c9ce64d6e5b. Feel free to try it out

Exactly the same for me. I had no clue about NLP or ML in 2018. CS224n together with CS231n kickstarted my learning journey and changed my life. I watched all lectures w/ notes, solved the problem sets with a stranger I met on the subreddit, and did a project on fake questions

🛠️ Weekend hack: Built EmbedKit to reduce re-work switching between multimodal embedding providers Same API for Cohere, ColPali, etc. Works with text, images, and PDFs. pip install embedkit Wdyt? What providers should I add next? 👀 github.com/databyjp/embed…

Came on to report something to 🏳️‍⚧️Graph Crimes🏳️‍🌈 Look at this drone shot of a 3-D pie chart. en.m.wikipedia.org/wiki/Kim_(Kore…

hey folks 👋🏻 what are the worst friction points in Hugging Face ecosystem (transformers, Hub etc) when you're building vision/multimodal stuff? I'd love to hear and see if we can help 🙏🏻

Is Siri dead?

This maths is driving me bananas. Does she then cease to exist in for the time that the kettle takes to boil?

I always get nice messages from ML folk about their latest research and try to amplify them 🤗 if you're a researcher and have something cool, post below here and I'll reshare 🔁