I am using the GPTWeaviateIndex combined with a custom llm which I have defined in my service_context. When I am attempting to build the WeaviateIndex it still errors on the fact that I have to provide an openai API key? Am I misunderstanding how the weaviateindex works? I assumed that the fact that it has been embedded by Weaviate would be enough for an index to be created, but it turns out that OpenAI is still required for something (?)
you'll either need to provide an embed_model as well in the service context (as that defaults to openai embeddings), OR you can just set a fake openai key if you aren't generating embeddings
