Vector Table
A vector table is where you store the records that you want to search. Autonomous AI Vector Database supports the creation of two types of vector tables; Integrated Embedding and Bring Your Own Vectors.
Integrated Embedding
Use the Integrated Embedding option when you would like the database to automatically generate vectors for you.
You can choose this option using the following process:
-
Load an ONNX embedding model.
-
Create a vector table and select the loaded embedding model.
-
Specify the Embedding metadata JSON path. This is the metadata field that contains the content that should be embedded.
-
Upsert records that include metadata with the metadata JSON path field. The database automatically generates the embedding using the loaded embedding model and stores it with the record.
Bring Your Own Vectors
Use the Bring Your Own Vectors option when you already generate embeddings outside the database or when you need full control over the embedding pipeline.
Choose this option by using the following steps:
-
Create a vector table with a dense vector type.
-
Generate vectors externally using your own embedding model.
-
Upsert records that include the vector values, plus any metadata that you want to filter on.
To get started quickly, the Vector Database Console also includes a Sample Data option that creates a vector table pre-populated with a National Parks sample data set. This is a convenient way to explore table creation, ingestion, and querying without preparing your own data first.