Create 微調整済み モデル接続

Create 微調整済み モデル接続 by customizing the supported foundational models via testing and fine-tuning to make them available to Pro Developers for connecting to the AI スキル they create.

Connect to your instances of the foundational models from hyperscaler vendors such as Amazon Bedrock, Google Vertex AI, Azure OpenAI, or OpenAI and customize their models by fine-tuning them and saving them with a specific name.

The Automation Admin creates and tests the 微調整済み モデル接続 and makes these available to the Pro Developers, who can connect to them when creating AI スキル. The モデル接続 are used in AI スキル to send prompts and receive responses from the models.

The Automation Admin creates custom roles and assigns these 微調整済み モデル接続 to the custom roles that are assigned to users to enable their access to the モデル接続.

This feature gives you the option to create your own custom models based on your specific use case. As these 微調整済み モデル接続 are customized and trained using your data, within your organization's environment, you can govern and monitor their use as per your compliance and governance policies.

Additionally, you can use the 微調整済み models created with Amazon Bedrock in AWS Services, and Google Vertex AI in Google データ ソース when creating 微調整済み モデル接続 in オートメーション・エニウェア.

Create 微調整済み モデル接続

A few things to consider

To use the 微調整済み モデル接続 you would first create them by via tooling provided by the hyperscaler models. These 微調整済み models are supported out-of-the-box by AI Agent Studio in オートメーション・エニウェア.

Refer to the following information to create your custom, 微調整済み models for each foundational model:

Amazon Bedrock:
  1. Create your 微調整済み model in Amazon Bedrock Services. See Fine-tuned models for Amazon Bedrock.
  2. Next, click Custom Models.
  3. Find your 微調整済み model and select it.
  4. Make a note of the Region where the model is deployed.
  5. Use the Model ARN and Region values procured from Amazon Bedrock to connect to your 微調整済み model when creating a モデル接続 in オートメーション・エニウェア.
Google Vertex AI:
  1. Create a custom model in Google Vertex AI. See How to create Fine-tuned models for Google Vertex AI.
  2. Next, navigate to Vertex AI > Model Garden.
  3. Click View My Endpoints & Models.
  4. You should see the ModelID, and the relevant Region values for your 微調整済み model.
  5. Procure the ProjectId for this model from the first page after you log into Google Vertex AI. You can also get it from the top left corner after you log in, where it is additionally displayed.
  6. Use the procured ModelID and Region values to connect to your 微調整済み model when creating a モデル接続 in オートメーション・エニウェア.
Azure OpenAI:
  1. To create a custom model in Azure OpenAI, see Customize a model with fine-tuning.
  2. Then create a deployment using this custom model. Make a note of the Model name and Deployment name.
  3. Use the Model name and Deployment name values to connect to your 微調整済み model when creating a モデル接続 in オートメーション・エニウェア.
OpenAI:
  1. Create your 微調整済み model in OpenAI. See Create a Fine-tuned model with OpenAI.
  2. Procure the Model ID that was generated for the 微調整済み model.
  3. Use this Model ID value when creating a モデル接続 in オートメーション・エニウェア.

前提条件

The Automation Admin creates the モデル接続 and requires these roles and permissions to manage モデル接続 for their business organization.
  • Role: AAE_Basic, Automation Admin custom role
  • Permission: Attended Bot Runner
  • Settings: AI Data Management must be enabled by the Automation Admin and the check box selected for Allow users to disable logs on AI Skills. Allow users with the Bot Creator license to disable data logging when using AI Skills.

See AI ツール のロールと権限 for the Automation Admin custom role permissions.

Other requirements:
  • You would first create a 微調整済み モデル接続 and save it with a specific name so that you can use it when creating a モデル接続. Refer to the section above to see how to create 微調整済み モデル接続 for each foundational model vendor.
  • If you want to store authentication details in a credential vault, have that information handy. See Credential Vault による安全な資格情報ストア.
  • To test a モデル接続, you must be connected to a Bot エージェント 22.60.10 and later. As part of the test, you would have to run the Bot on your desktop. Hence ensure the Bot エージェント is configured to your user. For this task, if you have to switch connection to a different Control Room, see: Control Room インスタンス間でのデバイス登録の切り替え.
  • You need access to the レコーダー package and the AI スキル package to test the connection successfully. A test プロンプト would be executed to test the モデル接続.

Follow these steps to create a 微調整済み モデル接続.

手順

  1. In your Control Room environment, navigate to AI > Model connections > Create model connection.
  2. In the Create model connection screen you would configure these Connection settings:
    1. Model connection name: Provide a name for easy identification of the モデル接続.
    2. Description (optional): Add a meaningful short description defining the connection.
    3. Choose a vendor: Choose a foundational model vendor from the supported list of vendors such as Amazon Bedrock, Google Vertex AI, Azure OpenAI, or OpenAI.
    4. Choose a type: Choose 微調整済み from the drop-down list.
    5. Choose a model: Choose a model from the drop-down list. The list displays relevant models as per your vendor selection.
      We recommend not selecting any of the Amazon Anthropic Claude models from the drop-down list. Amazon does not allow fine-tuning any of the Anthropic Claude models, hence this model option should not display in the drop-down list. We will remove the option in our future release.
    6. Fine-tuned model name: Enter the specific name of the 微調整済み custom model you created earlier.
    7. Click Next to proceed to the Authentication details section.
  3. In the Authentication details section, configure the settings as per values procured from the foundational model services for Amazon Bedrock, Google Vertex AI, Azure OpenAI, or OpenAI.
    注: For details on setting up the Authentication details for each model vendor, see モデル接続の認証.
  4. Click Test connection to make sure all connection details have been defined correctly and check if the connection is working.
    This is a desktop operation using a Bot エージェント. Use Bot エージェント 22.60.10 and later for successful testing.
    • If the connection works as expected, the system will process the request and you will get a system generated success message.
    • If the connection does not work as expected, you will get a system-generated message stating the reason for the connection failure. For example, if you have not downloaded the supported foundational model package to your workspace, you would get an error message. You would have to download the package and then retest the モデル接続.
    • If the testing a モデル接続 is unsuccessful or if you leave the task incomplete, the モデル接続 will not get saved and you will have to restart the process of creating the モデル接続.
  5. Click Next to proceed to the Invite roles section to begin assigning custom roles to users.
    The Automation Admin would create custom roles and assign the モデル接続 to the role, which can then be assigned to users. Only users assigned to this custom role can use this モデル接続.
  6. Assign access to the Pro Developer via custom role (using RBAC), for using this モデル接続 to create an AI スキル.
  7. Click Create model connection to complete creating the モデル接続.
    After successfully creating the モデル接続, the Pro Developer would use it to create an AI スキル

    See: AI スキルを作成.

次のステップ

After creating and testing the モデル接続, you would assign it to the Pro Developers, who would use this connection to create AI スキル.

See AI スキルを作成.

注: When you create or test an AI スキル in the AI Skill editor, the success or failure details along with the model responses can be viewed in these navigation screens:
  • Administration > AI governance > AI Prompt log
  • Administration > AI governance > Event log
  • Administration > Audit log

See AI ガバナンス.

As the next step in your sequence of tasks, go to AI スキルを作成 and create an AI スキル and connect to a 微調整済み モデル接続 to eventually use it in an automation.