Use LLMs to analyze wine reviews and test vector databases.
Example code used in the article: Raise a roast to the Vector Databases
This repo contains code that scrapes wine reviews from Wine Spectator magazine and then uses a sentence embadding model to embed the tasting notes. The resulting dataset of wines with their embeddings allow us to perform semantic (Vector) search and find wines with characteristics similar to a preferred wine.
A sample PCA analysis on the vectors shows how Wines from different varieties of grapes are closely clustered together:
