Fine-tuned foundation models

Foundation model tuning is at the Preview stage.

Yandex DataSphere allows you to tune foundation models on your own examples to make them more tailored to your specific needs. Tuning is based on the Fine-tuning method. You can send requests to a fine-tuned model through the DataSphere Playground interface or through the API using code.

Information about fine-tuned foundation models as a resourceInformation about fine-tuned foundation models as a resource

The following information is stored about each fine-tuned model:

• Name and description.
• Status.
• Foundation model.
• Learning rate.
• Name of the user who tuned the model.
• Data required for tuning.
• Model tuning date in UTC format, e.g., November 2, 2023, 17:39.

To get information about models tuned in a project:

1. Select the relevant project in your community or on the DataSphere homepage in the Recent projects tab.

2. In the list of available project resources, select Models.
3. In the Project tab, select Tuned foundation models.

Visibility scope of fine-tuned foundation modelsVisibility scope of fine-tuned foundation models

A fine-tuned model is a project resource. A community administrator can share fine-tuned models with other community projects by granting access to them in the Access tab on the model viewing page.

See alsoSee also

• Text model tuning
• Foundation models