jclee967 / openai-embeddings

openai-embeddings

This is a demonstration of using OpenAI API to run neural search based on vector embeddings.

• A sample data of product descriptions was downloaded from: https://www.kaggle.com/datasets/cclark/product-item-data . Data in a similar format can be used as well.
• First, documents (products descriptions in this case) are embedded into a vector space (generate_documents_embeddings.py)
• Next, one can run queries on the embeddings (query.py)

The model used is OpenAI's text-embedding-ada-002, which was released on December 15, 2022 (Release Notes).

Setup

• OpenAI API Key is required. Create a .env file (use .env.example as a template), and enter your API Key there.