Michael Hunger
Michael Hunger
Data importer/workspace can also connect to local neo4j instances. Good point on integrating the structured data, we'll think about it.
Added an enhancement here: https://github.com/neo4j-labs/llm-graph-builder/issues/596
Yeah I know this is frustrating, we might need to use special `__Document__`, and `__Chunk__` labels in the future.
That's something you can do when using the langchain integration directly. Not yet in the app, and no near term plans as it needs much additional metadata. https://python.langchain.com/v0.1/docs/use_cases/graph/constructing/#llm-graph-transformer
Good question, we would have to investigate, if you got time on looking at how to configure caching and contribute it, that would be great.
@chinmaydeshpandecaxsol can you share the example file?
in local deployment / DEV you can use "graph" search You need a clean schema though, that's why it's not enabled in the publicly deployed version.
Otherwise you'd have to add the structural elements that the app requires to make it work, e.g. Document node with fileName and status and Chunk nodes with text and embedding...
Across all LLMs, just total token usage.