This minimal notebook illustrates a very simple tour of Weaviate V4 python client.
- It depends on a local Weaviate container for which a docker-compose has been provided.
- You should run it either locally or via an SSH tunnel on the 8077 and 50051 ports.
- It also depends on your OpenAI API key being provided (in the
.envfile).- Copy/rename the
.env-examplefile to a.envfile.
- Copy/rename the
- The data directory contains description data about approximately 2000 ebooks. Most are English but there also are some Italian books.
- First of all, run
poetry installto add the required python packages. - Just run the notebook mybooks in the booksimil directory step by step observing its behaviors.
Note: The current version will throw an error if the data insertion portion is run again with the collection already created.
- To run all again either change the collection name in the
.envfile or delete the collection in the Weaviate database.
This program will read a collection of metadata about approx. 2000 books, will create a Weaviate collection for it using OpenAI's text2vec "text-embedding-ada-002" model. The rest of the notebook will show a few different types of semantic and generative queries.