VectorStoreIndex.from_documents, you can simply call the from_documents method again with the new documents and the same URI. Here's an example:index = VectorStoreIndex.from_documents(
[Document(text="The sky is blue")], uri="/tmp/new_dataset"
)
# Add more documents
index = VectorStoreIndex.from_documents(new_documents, uri="/tmp/new_dataset")new_documents is the list of new documents you want to add to the index. The uri parameter should be the same as the one you used when creating the index initially.Note: you can ask me a follow up question by @ mentioning me again :speech_balloon:insert method of the index to add documents one at a time. Here's an example:from llama_index import VectorStoreIndex
index = VectorStoreIndex([])
for doc in documents:
index.insert(doc)documents is a list of Document objects that you want to add to the index. This method allows you to add documents to the index one at a time, which can be useful if you are processing documents in a stream or if you want to add documents to the index as they become available.Note: you can ask me a follow up question by @ mentioning me again :speech_balloon:from llama_index import VectorStoreIndex, Document # Load the index from storage index = VectorStoreIndex.load_index_from_storage(uri="/path/to/index") # Create a new document doc = Document(text="New document text", id_="new_doc_id") # Insert the new document into the index index.insert(doc)
"/path/to/index" with the path to your index. The Document constructor is used to create a new document, and the insert method is used to add this document to the index.load_index_from_storage is not mentioned in the provided extracts, so please refer to the LlamaIndex documentation or source code for the correct method.Note: you can ask me a follow up question by @ mentioning me again :speech_balloon:service_context into the insert method of an index. The insert method is typically used to add a document to an index, and it's not clear from the provided information whether it accepts a service_context as an argument.service_context is used when creating an index or loading an index from storage. For example:from llama_index import StorageContext, load_index_from_storage, ServiceContext # rebuild storage context storage_context = StorageContext.from_defaults(persist_dir="<persist_dir>") # define service context service_context = ServiceContext.from_defaults() # load index index = load_index_from_storage(storage_context, service_context=service_context)
service_context is passed as an argument to the load_index_from_storage function. The service_context contains various settings and helpers that are used by the index.service_context can be passed to the insert method, please refer to the LlamaIndex documentation ().Note: you can ask me a follow up question by @ mentioning me again :speech_balloon:VectorStoreIndex, you can specify the service_context for each index individually when you create them. Here's an example:from llama_index import VectorStoreIndex, ServiceContext
# Define the service context
service_context = ServiceContext.from_defaults()
# Create a list of indices
indices = []
for i in range(5):
index = VectorStoreIndex.from_documents(documents, service_context=service_context)
indices.append(index)documents is a list of documents that you want to add to each index. The service_context is defined once and then used for each index. This ensures that all indices use the same service context.Note: you can ask me a follow up question by @ mentioning me again :speech_balloon: