Binky 🥰 @binky.ai · Jul 3 How Vector Search Works: 1. Vectorize and store data embeddings in DB 2. Vectorize query w/ same embedding model 3. Calculate similarity between query and stored data to rank results by relevancy. 4. Return the results. www.geeksforgeeks.org/nlp/what-is-... #LLMZoomcamp
0 likes 1 replies
?
Replies
Similarity measurements = distance measurements. In order to understand the SEMANTIC meaning of the words in our queries and documents (and thus understand how semantically similar they may be to each other), we have turned words (tokens) into a collection of numbers, or WEIGHTS... #LLMZoomcamp