Knowledge graph

We do have a knowledge graph index. You can manually insert existing triplets (rather than the default of letting the LLM extract them), but each triplet also needs a node that has the text where that triplet came from

Although you could work around this by passing in dummy nodes, and then using include_text=False when querying

https://gpt-index.readthedocs.io/en/latest/examples/index_structs/knowledge_graph/KnowledgeGraphDemo.html#id1

Thanks @Logan M! Our company also has its own strucutral causal models (SCMs) in DAG of variety of causal variables and mechanism (a functional relationship) between them - is there any way that I can use those SCMs to inject them as a context using llama-index?

Not sure, I think you lost me a bit there haha. Might need an example of what these SCMs do