Create your custom embedding model in the Generative AI Model Configuration table so that your AI Search RAG application can use it to generate embeddings for semantic indexing. This setup ensures the model is recognized and properly connected to send and receive requests.

Vorbereitungen

You must create a connection and credential alias for your embedding model. For more information, see Create a Connection & Credential alias

Role required: admin

Prozedur

1. Navigate to All, and then enter sys_generative_ai_model_config.list in the filter to go to the Generative AI Model Configuration [sys_generative_ai_model_config] table.
2. Select New.
3. On the form, fill in the fields.

Tabelle : 1. Generative AI Model Configuration

Field	Description
Active	Option to activate the embedding model.
Model	A unique name for your embedding model.
Domain	Domain you want to associate the model with. For example, AI Search RAG.
External	Option to make this model to be used external.
Connection and Credential Alias	Connection and credential alias that you created on your own for the custom embedding model.
Supported Language	Languages supported for this model. By default, the supported language is English.
Model Type	The type of model used for a specific purpose. For example Embedding Model.
Vector Dimension	The value shouldn't exceed 4096. This field appears only if you have selected Embedding Model in the Model Type field.
Application	Name of the application this model is configured for.
Provider	Name of the Generative AI provider mapping.
Max Tokens	Maximum limit to generate embeddings.

4. Select Submit.

Ergebnisse

The new embedding model is created.

Nächste Maßnahme

Set a provider for an embedding model.