Find answers from the community

Updated 6 months ago

Weaviate

At a glance

The community member asks if the llama index supports custom vectorIndexConfig parameters, such as efConstruction, when using Weaviate. Another community member responds that the llama index accepts a few parameters, and that the user can optionally initialize the index with a custom schema, but some parts need to be the same for querying to work. The community member then states they would like to change efConstruction and possibly other parameters, and another community member confirms that the user will need to create the schema ahead of time, providing a link to the default schema.

Useful resources
Hi everyone. When using weaviate, does llama index support custom vectorIndexConfig parameters such as efConstruction, either by specifying them directly, or by using an existing schema during index construction, rather than creating a new one?
L
M
4 comments
Sorry, what parameters did you want to change?

Right now we accept a few parameters. We create a default schema, but you can optionally initialize the index with a custom schema, but some parts need to be the same on order for querying to work

Looking at the source code here my help
https://github.com/jerryjliu/llama_index/blob/b0f893c696b3aec20d4304c80c19f8a8fc692f85/llama_index/vector_stores/weaviate.py#L67
I'd like to change efConstruction and possibly other parameters described here:
https://weaviate.io/developers/weaviate/configuration/indexes
Ah I see. Ya you'll have to create the schema ahead of time then

For reference, this is the default schema
https://github.com/jerryjliu/llama_index/blob/b0f893c696b3aec20d4304c80c19f8a8fc692f85/llama_index/vector_stores/weaviate_utils.py#L82
Thank you, that's exactly what I needed.
Add a reply
Sign up and join the conversation on Discord