gk1

36 ms·

811.
by gk1 4y ago
Biggest difference is Milvus is self-hosted (manage your own infra) while Pinecone is a managed service. There are other differences but that’s often a driving factor. We (Pinecone) tend to attract customers who want to start, ship, and sca
812.
Building a Multi-User Chatbot(pinecone.io)
2 points by gk1 4y ago|0 comments
813.
Retrieval in LangChain(blog.langchain.dev)
212 points by gk1 4y ago|69 comments
814.
We built semantic search for ArXiv(sigmoidprime.com)
60 points by gk1 4y ago|3 comments
815.
by gk1 4y ago
The biggest deal about this is the ability to create your own plugins. The Retrieval Plugin is a kind of starter kit, with built-in integrations to the Pinecone vector database: https://github.com/openai/chatgpt-retriev
816.
by gk1 4y ago
100% agree. All the launch-partner apps (Kayak, OpenTable, etc) are there to grab attention but this plugin is the real big deal. It's going to let developers build their own plugins for ChatGPT that do what they want and access the
817.
by gk1 4y ago
https://github.com/openai/openai-cookbook/blob/main/examples... https://github.com/pinecone-io/examples/blob/master/generati... https://www.pinecone.io&#x2
818.
by gk1 4y ago
I think it's a combination of data, LLM quality, embedding search quality, and creativity.
819.
by gk1 4y ago
I'd say Microsoft. And they've been demonstrating that quite well.
820.
by gk1 4y ago
I suspect the author knows a lot more about how SWEs work than even many SWEs... https://en.wikipedia.org/wiki/Steve_Yegge
821.
by gk1 4y ago
I work for a vector database company (Pinecone) and can confirm that most of the mind-blowing built-with-ChatGPT products you see launching every eight'ish hours are using this technique that Steve describes. That is, embedding interna
822.
Show HN: Simple Sticky Bar – For when you just need a basic notification bar(github.com)
4 points by gk1 4y ago|0 comments
823.
by gk1 4y ago
You probably already know about Pinecone. If you have questions or want help testing anything let me know here or greg@pinecone.io
824.
Chatbot Memory with Langchain(pinecone.io)
4 points by gk1 4y ago|0 comments
825.
I fixed Notion with NLP to help get better search results(indiehackers.com)
1 points by gk1 4y ago|0 comments
826.
by gk1 4y ago
The first “recommended tweet” under the thread was some porn that I don’t even think is legal (hint: there was a dog involved). Posted 11 hours ago. I wonder what’s left of the content moderation team at Twitter.
827.
by gk1 4y ago
Moving through text using Cmd + Left/Right arrow will jump over two words if there’s an em dash between them with no spaces. As a frequent em dash user that was very annoying, so I switched to adding spaces — to hell with the APA.
828.
Image Recognition App in JavaScript(pinecone.io)
2 points by gk1 4y ago|0 comments
829.
Talk to Wikipedia Using ChatGPT(share.streamlit.io)
1 points by gk1 4y ago|1 comments
830.
Chatbase: Custom ChatGPT trained on your website(chatbase.co)
2 points by gk1 4y ago|1 comments
831.
Regularization in Neural Networks(pinecone.io)
2 points by gk1 4y ago|0 comments
832.
Welcome Vector Databases, Welcome Bob(pinecone.io)
16 points by gk1 4y ago|3 comments
833.
LangChain: The Unofficial Intro(pinecone.io)
2 points by gk1 4y ago|0 comments
834.
Image Recognition App in JavaScript Using Pinecone, Hugging Face, Vercel(pinecone.io)
1 points by gk1 4y ago|0 comments
835.
by gk1 4y ago
We'll participate in BigANN this year or next. It's a question of engineering/research resources.
836.
by gk1 4y ago
Correction: I'm not the Founder, but I've been at Pinecone for two years, from when we were ~10 people. Long enough to know what's under the hood. I've also been on HN long enough to expect a few cynics and conspiracy th
837.
by gk1 4y ago
You might be surprised to learn there’s no Faiss inside Pinecone. All the indexes are proprietary and have features you won’t find in Faiss, such as metadata filtering, real-time index updates, and horizontal/vertical scaling. Oh and o
838.
Support for sparse-dense embeddings for better search results(pinecone.io)
1 points by gk1 4y ago|0 comments
839.
by gk1 4y ago
From the article: > The embeddings were then indexed with Pinecone.
840.
SPLADE for Sparse Vector Search Explained(pinecone.io)
52 points by gk1 4y ago|14 comments

More ›