Expanded Self-Query Retriever 👀 and Self-querying with MyScale 🤩

[mpskex]

Expanded Self-Query Retriever 👀 and Self-querying with MyScale 🤩

LangChain is built with a great interface for self-query retrievers and many super nice contribution has been made during the last few weeks. 🌟 Self-query retrievers are really cool and we believe it will be the future of vector databases! 😎 So we decided to expand self-query retriever with more functions and data types enabled and bring MyScale to the big family of self-query enabled vector databases! 🤗

MyScale can make use of various data types and functions for filters. It will boost up your LLM app no matter if you are scaling up your data or expand your system to broader application. 🚀🚀🚀 And we believe it will be a common feature to every vector database in the future!

In the notebook we'll demo the SelfQueryRetriever wrapped around a MyScale vector store with some extra piece we contributed to LangChain. In short, it can be concluded into 4 points:

1. Add contain comparator 🪣 to match list of any if there is more than one element matched
2. Add timestamp data type 🕰️ for date time match (ISO-format, or YYYY-MM-DD)
3. Add like comparator 🔎 for string pattern search
4. Add arbitrary function 💣 capability.

Let's boost LLM apps with vector databases! 🎉🎉🎉

Before submitting

• unit-test for LIKE and CONTAIN
• Notebook for MyScale Self-query Retriever

Who can review?

Tag maintainers/contributors who might be interested:

• @hwchase17
• @dev2049