Architecture

Vector Database

Définition

A specialized database designed to store and efficiently search high-dimensional vectors (embeddings). Vector databases are essential for RAG systems, similarity search, and recommendation engines.

Exemples

Pinecone
Weaviate
Milvus
pgvector

Termes Connexes

Embedding

A numerical representation of data (text, images, etc.) in a multi-dimensional vector space. Embeddings capture semantic meaning, allowing similar concepts to be mathematically close together.

Retrieval-Augmented Generation (RAG)

A technique that enhances LLM responses by retrieving relevant information from external knowledge sources before generating a response. RAG improves accuracy, reduces hallucinations, and enables LLMs to access up-to-date information.

Services Connexes

Production Ai Systems

Besoin d'Aide pour Comprendre l'IA?

Réservez une consultation pour discuter de l'application des concepts IA à vos défis.