vector-search-elastic

This is the repository for all the material on Text Embeddings and Vector Search with Elasticsearch and Open-Source Technologies.

For a step-by-step description read our blog posts:

Elasticsearch Neural Search Tutorial

Elasticsearch Neural Search Tutorial (Platinum/Enterprise)

Elasticsearch Neural Search Improvements in 8.6 and 8.7

Requirements:

To replicate this work just install the requirements.txt in your python environment. e.g. (Python 3.8)

using pip

pip install -r requirements.txt

Repository content

• from_text_to_vectors: contains the python script to generate vector embeddings from MS Marco data
  • example_input: contains the MS Marco data (10k)
  • example_output: contains the vector embeddings obtained (10k)
• indexing_phase: contains the python scripts to index batches of documents to Elasticsearch at once from a file
• nlp_models: contains the import_model.py python script to import the all-MiniLM-L6-v2 sentence transformer from HuggingFace to Elasticsearch

Pipeline:

To run Elasticsearch (after downloading it):

cd elasticsearch-8.8.0
bin/elasticsearch
curl localhost:9200

To produce vectors externally:

cd from_text_to_vectors
python batch-sentence-transformers.py "./example_input/documents_10k.tsv" "./example_output/vector_documents_10k_384.tsv"

To index batches of documents to Elasticsearch:

cd indexing_phase
python indexer_elastic.py "../from_text_to_vectors/example_input/documents_10k.tsv" "../from_text_to_vectors/example_output/vector_documents_10k_384.tsv" "../from_text_to_vectors/example_output/vector_documents_10k_768.tsv"

To transform a query into vectors:

cd from_text_to_vectors
python single-sentence-transformers.py

Advanced feature (Platinum or Enterprise License)

If you run import_model.py with basic license you got the following error:

elasticsearch.AuthorizationException: AuthorizationException(403, 'security_exception', 'current license is non-compliant for [ml]')

To use it, start a free trial:

curl -XPOST http://localhost:9200/_license/start_trial?acknowledge=true

To import and load a language model to do inference directly within Elasticsearch:

cd nlp_models
python import_model.py

To index batches of documents to Elasticsearch using a Text Embedding Ingest Pipeline:

cd indexing_phase
python indexer_elastic_with_pipeline.py "../from_text_to_vectors/example_input/documents_10k.tsv"