#EXAMPLE TO STREAM THE DATASET

script.py

from datasets import load_dataset
wiki_data = load_dataset(
       'vblagoje/wikipedia_snippets_streamed',
        split='train',
        streaming=True
).shuffle(seed=none)

for d in wiki_data:
        print(d["article_title"])

#  ETL pipeline: extract > transform > charge in vector DB

for batch in wiki_data.iter(batch_size=1000):
        embeddings = model.encode(batch["passage_text"])
        index.upsert(embeddings)