🗡️LanceDB Vectorstore integration
@lancedb is a serverless, production-scale vector search that is fully open source
You can now use it within langchain!
Thank Minh Le for adding!
Docs: https://t.co/UAUMg0YAyf
Windows support has landed in Lance! We're excited to serve the windows dev community.
Also, I wanted to take the opportunity to express gratitude to Christoph Gohlke and the many unsung heroes of OSS (yes there's an xkcd for that).
https://t.co/RDuTLbYYXn
@RealRichomie@pinecone@milvusio@gpt_index Hey, @RealRichomie these are great insights! We are building a disk-based vector indices in Lance project (https://t.co/IwQ0I6eryX). Also, DiskANN is another disk/ssd based solution. Would love to hear your thoughts on them.
In this post we peel back the covers and build a Q&A app to search YouTube videos. We do it with Lance + basic OSS python tooling + OpenAI API. In this workflow, Lance serves as the vector index + document store.
https://t.co/5fHGbeAzjl
If you are in SF this eve swing by our AI Mixer at the @SieveData office at 5. AI Founders, Practitioners, and Enthusiasts welcome. Co-hosting with
@AI__Pub@EtodotAI
https://t.co/sohunfcgiy
@sievedata@etodotai
1 of 10 Pandas 2.0 is coming soon! But what's new? What are the breaking changes? Do I need to learn even more pandas APIs? Not to worry, Lance and @langchain has you covered https://t.co/Sp4mNhCNVT
Apologies to anyone who came to Lance github via Twitter yesterday. We weren't ready to announce yet but the cover was blown. We're rushing to get the docs back in order now!
https://t.co/W8briE5nQH
If you work with ML data (vectors, images, videos, and of course tabular data) using any @ApacheArrow compatible interface (e.g., @duckdb, @pandas_dev, @raydistributed), dataset versioning is now *free* with Lance (1/n)
https://t.co/HRyhlwO4Ho
OK so this was a TON of fun: we wrote an extension for @duckdb to run @PyTorch models (powered by our @ApacheArrow compatible Lance data format). I’m calling it…drum roll please…peeking duck (i’ll be here all night folks).
https://t.co/UbhZo4YLqh
Curious what we are working on? Here is the first blog of a series: Active learning made simple https://t.co/nhiOcZ6caw #rikai#ActiveLearning#MachineLearning