A brief description of your project goes here.

To retrieve data use --retrieve flag and choose the correct name of the database with --db_name. The nokia_rag database is already setup for vector search and ready for querying. Please use this database name only for execution of the below command as the code runs with admin permissions. python main.py --retrieve --setup database_medium --db_name nokia_rag

Data preparation To prepare the data for the document database python main.py --setup data --load_csv_source_filename "example.csv" --save_json_source_filename "example_org.json" --save_json_embeddings_filename "example_chunked_embeddings.json"

The "example.csv" file is small just to show that the code runs. To run the code on the medium articles dataset set "medium.csv" as the source filename.

Note: The the right filenams need to be used when launching the database.

To launch the database for the first time (done once per database). Inserting the data into the database. python main.py --launch --setup database_medium --db_name <unsearchable_demo_database_name> --medium_json_filename "example_org.json" --embeddings_json_filename "example_chunked_embeddings.json"

Note: Unfortunately for the free cluster tier of the atlas databse I chose it is not possible to set search index in the database from code, so the database setup cannot be fulle reproduced with just this repo. Some interaction with UI is necessary. For this reason for retrieval use already set up nokia_rag database.

Use pytest command, but the test are likely broken as for now...