Vector search

Embed a corpus and find the most similar sentence via cosine similarity.

1. Embed your corpus

   search.py

   from ai_sdk import openai, embed_many, cosine_similarity
   
   sentences = [
   "The cat sat on the mat.",
   "A dog was lying on the rug.",
   "Paris is the capital of France.",
   "Cats love sunny spots.",
   ]
   model = openai.embedding("text-embedding-3-small")
   res = embed_many(model=model, values=sentences)
   
   

2. Search for a query

   query = "Dogs enjoy napping on carpets."
   q_emb = embed_many(model=model, values=[query]).embeddings[0]
   
   scores = [cosine_similarity(q_emb, emb) for emb in res.embeddings]
   most_similar_idx = max(range(len(scores)), key=scores.__getitem__)
   print("Most similar: ", sentences[most_similar_idx])
   

   Expected result:

   Most similar:  A dog was lying on the rug.
   

cosine_similarity is dependency-free. Replace it with NumPy or faiss in production for better performance.