L4 why does index give a different number of records/docs vs other methods?

why does index give a different number of records/docs?


Loader gives 1000 (as we know, 1000 rows):

docs = loader.load()
len(docs)

Index gives 1210:

index = VectorstoreIndexCreator(
    vectorstore_cls=DocArrayInMemorySearch
).from_loaders([loader])

index.vectorstore.doc_index.num_docs()

but db gives 1000 (like the number of rows):

db = DocArrayInMemorySearch.from_documents(
    docs, 
    embeddings
)
db.doc_index.num_docs()

and retriever gives 1000 (as I would expect since it’s using db):

retriever = db.as_retriever()
retriever.vectorstore.doc_index.num_docs()