Managing AI settings [Technical Preview]
Cloudera Data Visualization offers extensive site-wide configuration options, enabling users with administrative privileges to manage various settings crucial for organizational workflows. You can enable the AI visual, select the AI engine, and specify additional settings.
General
In this section, you can enable the AI visual. Before enabling this feature, make sure you read the legal terms and conditions provided in the Info tooltip.
When you enable the feature for the first time, a modal window prompts you to accept the terms and conditions for using the AI visual. The system records the date and the username of the individual accepting the terms.
Additionally, you can also enable the redaction of AI visual logs. If enabled, audit logs are redacted to omit information transmitted by the AI visual.
Embeddings
In this section, you can manage embedding profiles. You can create new profiles, view the list of existing ones, and perform actions such as copying, editing, deleting existing profiles, or setting a new default profile.
Creating a new embeddings profile
-
Click CREATE PROFILE.
-
Provide a name for the profile.
-
Select the AI engine you want to use for the embeddings.
You have three options: OpenAI, MiniLM, and Other
-
Specify the settings for the selected AI engine.
-
API key for the embeddings model: required to access the external third-party AI service provider
Ensure you read the legal terms and conditions provided in the Info tooltip.
-
Model: used for the embeddings
-
API key for the embeddings model: required to access the external third-party AI service provider
Ensure you read the legal terms and conditions provided in the Info tooltip.
-
Service URL: URL for the embedding service
-
Service input parameter name: name for the embedding service input
-
Service response parameter name: name for the embedding service response parameter
-
API key for embeddings model: required to access the external third-party AI service provider
Ensure you read the legal terms and conditions provided in the Info tooltip.
-
Service URL: URL for the embedding service
- AI engine settings: use this textbox to provide your custom request and response transformer functions in valid Python.
-
Cloning an existing embeddings profile
Alternatively, you can clone an existing embeddings profile to create a new profile instance.
-
Click to open the Create Profile modal. All details are pre-filled from the original profile, except for the Name field.
- Enter a new name for the profile.
Managing embeddings profiles
-
Setting a default profile: The current default profile that will be used for new AI visuals is marked. To change the default, hover over a non-default profile and click the Set as default option.
-
Deleting a profile: Click on the profile row.
-
Editing a profile:
- Click on the profile row to open the Edit Profile modal window.
- Make the necessary changes to the profile.
- Click SAVE.
Completion
In this section, you can manage completion profiles. You can create new profiles, view the list of existing ones, and perform actions such as copying, editing, deleting existing profiles, or setting a new default profile.
Creating a new completion profile
-
Click CREATE PROFILE.
-
Provide a name for the profile.
-
Select the AI engine you want to use for the completion.
You have three options: OpenAI, Llama, and Other
-
Specify the settings for the selected AI engine.
-
API key for the completion model: required to access the external third-party AI service provider
Ensure you read the legal terms and conditions provided in the Info tooltip.
-
Model: used for the embeddings
-
Temperature: controls the randomness of the output on a scale from 0 to 2Lower values produce more focused and deterministic responses, while higher values introduce greater randomness and creativity. The default setting is 1.
-
API key for the completion model: required to access the external third-party AI service provider
Ensure you read the legal terms and conditions provided in the Info tooltip.
-
Service URL: URL for the completion service
-
Service input parameter name: name for the completion service input
-
Service response parameter name: name for the completion service response parameter
-
Extra arguments
-
API key for completion model: required to access the external third-party AI service provider
Ensure you read the legal terms and conditions provided in the Info tooltip.
-
Service URL: URL for the completion service
- AI engine settings: use this textbox to provide your custom request and response transformer functions in valid Python.
-
Cloning an existing completion profile
Alternatively, you can clone an existing completion profile to create a new profile instance.
-
Click to open the Create Profile modal. All details are pre-filled from the original profile, except for the Name field.
-
Enter a new name for the profile.
Managing completion profiles
-
Setting a default profile: The current default profile that will be used for new AI visuals is marked. To change the default, hover over a non-default profile and click the Set as default option.
-
Deleting a profile: Click on the profile row.
-
Editing a profile:
- Click on the profile row to open the Edit Profile modal window.
- Make the necessary changes to the profile.
- Click SAVE.