The community members are discussing the behavior of the VectorStoreIndex.from_documents() function. The original poster asks if manually adding embeddings to a document will cause the function to generate those embeddings again, and if the embeddings will be saved to the storage context.
The comments indicate that the function will generate embeddings for the test_docs, but there is some uncertainty around this. One community member suggests that the function does not generate any embeddings, based on their own testing. Another community member suggests that the function can create an index without any content by using an empty list, and that the number of test_docs may be different from the number of nodes.
The community members discuss ways to ensure that a node has the same content as a document, and suggest using the node method instead of from_documents if the user does not care about uniform node sizes.
if i manually add embeddings for a document , will this code make those embeddings again? And will it save to the storage context? index = VectorStoreIndex.from_documents( test_docs, storage_context=storage_context, embed_model=embed_model, show_progress=True )
