Code Examples

from openai import OpenAI from veep import Client openai = OpenAI() vp = Client("your-api-key") # Embed and store documents documents = ["Vector databases store embeddings...", "RAG improves LLM accuracy..."] for i, doc in enumerate(documents): embedding = openai.embeddings.create( model="text-embedding-3-small", input=doc ).data[0].embedding vp.add(vectors=[embedding], ids=[str(i)], metadata=[{"text": doc}]) # Query: find relevant context for a question question = "How do vector databases work?" q_embedding = openai.embeddings.create( model="text-embedding-3-small", input=question ).data[0].embedding results = vp.search(vector=q_embedding, top_k=3) context = "\n".join(r.metadata["text"] for r in results) # Pass context to LLM response = openai.chat.completions.create( model="gpt-4", messages=[{"role": "user", "content": f"Context: {context}\n\nQuestion: {question}"}] )

from transformers import CLIPModel, CLIPProcessor from PIL import Image from veep import Client model = CLIPModel.from_pretrained("openai/clip-vit-base-patch32") processor = CLIPProcessor.from_pretrained("openai/clip-vit-base-patch32") vp = Client("your-api-key") # Index images for path in image_paths: image = Image.open(path) inputs = processor(images=image, return_tensors="pt") embedding = model.get_image_features(**inputs)[0].detach().numpy() vp.add(vectors=[embedding], ids=[path], metadata=[{"path": path}]) # Search by text: "a photo of a sunset over the ocean" text_inputs = processor(text=["a photo of a sunset over the ocean"], return_tensors="pt") query_vec = model.get_text_features(**text_inputs)[0].detach().numpy() results = vp.search(vector=query_vec, top_k=10) for r in results: print(f"{r.metadata['path']} score={r.score:.4f}")

from veep import Client vp = Client("your-api-key") # Index products with metadata vp.add( vectors=product_embeddings, ids=product_ids, metadata=[ {"name": "Running Shoe X", "category": "footwear", "price": 89.99}, {"name": "Trail Runner Pro", "category": "footwear", "price": 129.99}, # ... ] ) # Find similar products to what a user is viewing current_product_vector = get_product_embedding("product-123") similar = vp.search( vector=current_product_vector, top_k=6, filter={"category": "footwear"} ) for product in similar: print(f"{product.metadata['name']} ${product.metadata['price']}")

from veep import Client from sentence_transformers import SentenceTransformer model = SentenceTransformer("all-MiniLM-L6-v2") vp = Client("your-api-key") # Index all documents embeddings = model.encode(documents) vp.add(vectors=embeddings, ids=doc_ids) # Find duplicates: search each doc against the collection SIMILARITY_THRESHOLD = 0.95 duplicates = [] for i, emb in enumerate(embeddings): results = vp.search(vector=emb, top_k=5) for r in results: if r.id != doc_ids[i] and r.score > SIMILARITY_THRESHOLD: duplicates.append((doc_ids[i], r.id, r.score)) print(f"Found {len(duplicates)} duplicate pairs")