Azure AI Studio model catalog vectorizer

  • Article

Important

This vectorizer is in public preview under Supplemental Terms of Use. The 2024-05-01-Preview REST API supports this feature.

The Azure AI Studio model catalog vectorizer connects to an embedding model that was deployed via the Azure AI Studio model catalog to an Azure Machine Learning endpoint. Your data is processed in the Geo where your model is deployed.

Vectorizer parameters

Parameters are case-sensitive. Which parameters you choose to use depends on what authentication your AML online endpoint requires, if any.

Parameter name Description
uri (Required) The URI of the AML online endpoint to which the JSON payload is sent. Only the https URI scheme is allowed.
key (Required for key authentication) The key for the AML online endpoint.
resourceId (Required for token authentication). The Azure Resource Manager resource ID of the AML online endpoint. It should be in the format subscriptions/{guid}/resourceGroups/{resource-group-name}/Microsoft.MachineLearningServices/workspaces/{workspace-name}/onlineendpoints/{endpoint_name}.
region (Optional for token authentication). The region the AML online endpoint is deployed in. Needed if the region is different from the region of the search service.
timeout (Optional) When specified, indicates the timeout for the http client making the API call. It must be formatted as an XSD "dayTimeDuration" value (a restricted subset of an ISO 8601 duration value). For example, PT60S for 60 seconds. If not set, a default value of 30 seconds is chosen. The timeout can be set to a maximum of 230 seconds and a minimum of 1 second.

What authentication parameters to use

Which authentication parameters are required depends on what authentication your AML online endpoint uses, if any. AML online endpoints provide two authentication options:

  • Key-Based Authentication. A static key is provided to authenticate scoring requests from the vectorizer.
    • Use the uri and key parameters
  • Token-Based Authentication. The AML online endpoint is deployed using token based authentication. The Azure AI Search service's managed identity must be enabled. The vectorizer then uses the service's managed identity to authenticate against the AML online endpoint, with no static keys required. The identity must be assigned owner or contributor role.
    • Use the resourceId parameter.
    • If the search service is in a different region from the AML workspace, use the region parameter to set the region the AML online endpoint was deployed in

Supported vector query types

Which vector query types are supported by the AI Studio model catalog vectorizer depends on the modelName that is configured.

modelName Supports text query Supports imageUrl query Supports imageBinary query
OpenAI-CLIP-Image-Text-Embeddings-vit-base-patch32 X X X
OpenAI-CLIP-Image-Text-Embeddings-ViT-Large-Patch14-336 X X X
Cohere-embed-v3-multilingual X

Expected field dimensions

The expected field dimensions for a field configured with an AI Studio model catalog vectorizer depend on the modelName that is configured.

modelName Expected dimensions
OpenAI-CLIP-Image-Text-Embeddings-vit-base-patch32 512
OpenAI-CLIP-Image-Text-Embeddings-ViT-Large-Patch14-336 768
Cohere-embed-v3-english 1024
Cohere-embed-v3-multilingual 1024

Sample definition

"vectorizers": [
    {
        "name": "my-ai-studio-catalog-vectorizer",
        "kind": "aml",
        "amlParameters": {
            "uri": "https://my-aml-endpoint.chinaeast.inference.studio.ml.azure.cn/score",
            "key": "0000000000000000000000000000000000000",
            "timeout": "PT60S",
            "modelName": "OpenAI-CLIP-Image-Text-Embeddings-vit-base-patch3",
            "resourceId": null,
            "region": null,
        },
    }
]

See also