What is Ansible Lightspeed with IBM watsonx Code Assistant?¶
Ansible Lightspeed with IBM watsonx Code Assistant is a service to enhance the creation of Red Hat Ansible automation. Ansible Lightspeed with IBM watsonx Code Assistant uses, among other data, Ansible roles and collections that are available through the community website, Ansible Galaxy. The following provides some answers to questions you may have related specifically to Ansible Lightspeed with IBM watsonx Code Assistant.
Do you match Recommendations to Training Data?¶
Given the nature of deep learning technology, as well as the kinds of content used to train and which are generated, it is not possible to identify specific training data inputs that contributed to particular output recommendations. Nevertheless, Ansible Lightspeed with IBM watsonx Code Assistant includes a feature to help users that are interested in understanding possible origins of generated content recommendations. When Ansible Lightspeed with IBM watsonx Code Assistant generates a recommendation, it will attempt to find items in the training dataset that closely resemble the recommendation. In such cases, Ansible Lightspeed with IBM watsonx Code Assistant will display licensing information and a source repository link for the training data matches in a panel interface in the VS Code extension. Licensing information is determined using the GitHub license API.
This feature may enable users to ascertain open source license terms that are associated with related training data. We have implemented this feature even though we believe it is unlikely that either the training data used in fine-tuning or the output recommendations themselves are generally protected by copyright, or that output reproduces training data content controlled by copyright licensing terms.
Does Red Hat assert rights in the suggestions generated by the Ansible Lightspeed with IBM watsonx Code Assistant service?¶
Red Hat does not claim any copyright or other intellectual property rights in the suggestions generated by the Ansible Lightspeed with IBM watsonx Code Assistant service.
Is my final Ansible content created from the suggestions in VS Code editor used as feedback to the Ansible Lightspeed with IBM watsonx Code Assistant service and may I opt out of providing feedback?¶
Ansible Lightspeed with IBM watsonx Code Assistant is designed to be improved through feedback on the quality of its suggestions. Technical details of user experiences with Ansible Lightspeed with IBM watsonx Code Assistant can be useful in informing further improvements. For example, identifying which Ansible Lightspeed with IBM watsonx Code Assistant suggestions has been accepted by the user, noting particular details that the user has needed to adjust, etc.
This feedback is obtained by way of the Ansible Lightspeed with IBM watsonx Code Assistant VS Code extension and is used to improve the Ansible Lightspeed with IBM watsonx Code Assistant service.
Can Ansible Lightspeed with IBM watsonx Code Assistant generate Ansible Playbooks?¶
No. Ansible Lightspeed with IBM watsonx Code Assistant only generates Ansible tasks in Tech Preview.
How do I provide general feedback?¶
The Ansible VS Code extension links a sentiment survey which we encourage you to complete as you test the Ansible Lightspeed with IBM watsonx Code Assistant service. To access the survey, click “Lightspeed” on the right-hand side of the VS Code status bar and click “Take Survey”.
After completing the survey, you have the opportunity to provide general feedback. You also can
contact us in the Ansible Lightspeed with IBM watsonx Code Assistant Matrix room:
What information is sent from VS Code to train the model?¶
Red Hat attempts to remove or anonymize personal information in the telemetry data before being sent to train the model. The telemetry data includes, but is not limited to the following:
- A session and activity ID is used to correlate the telemetry data.
- Information on whether the recommendation was ignored, modified, or accepted.
- The Ansible playbook content in which the recommendation was generated. Potential sensitive information is first removed.
- The Ansible playbook file name.
More detail about the telemetry captured by the service is available here.
What data sources are used to train the IBM watsonx Code Assistant model?¶
The IBM watsonx Code Assistant model gathers training data from Ansible Galaxy, GitHub, and other open sources of data. Ansible Galaxy users can opt-out of their Ansible Galaxy namespaces that can be used to train the IBM watsonx Code Assistant model. Please refer to the Contributing section in this document for more information on Ansible Galaxy.
How long will it take before I see my feedback improving the model?¶
Training the IBM watsonx Code Assistant model with the telemetry data is resource-intensive. IBM Research intends to retrain the model at a cadence allowing noticeable model improvements between model versions.
How do we address potentially sensitive information in telemetry data?¶
The Ansible VS Code extension does not directly collect personal user information. Known sensitive information, such as identifiable personal information, credentials, or keys, that can be identified programmatically, is removed before being sent to telemetry storage. However, Ansible content allows for content developers to write comments, secrets, and other data that cannot always be captured and scrubbed programmatically. It is up to the end-user to ensure that best practices are followed to prevent sensitive information from leaking from your organization, including the use of variables to reduce the likelihood that sensitive data could exist within Ansible content.
How long does Red Hat keep telemetry?¶
Telemetry data may be kept indefinitely for product improvement.
Can I request a telemetry data purge?¶
Users are not able to request that their telemetry be purged. You may disable the Ansible Lightspeed service to prevent further telemetry from being sent to the service.
Can I get different responses for the same prompt?¶
In general, Ansible Lightspeed with IBM watsonx Code Assistant should provide the same response for the same prompt. This behavior may change over time.
I got an error "Too many requests to Ansible Lightspeed. Please try again after some time." How long do I need to wait?¶
The current per-user rate limit is 10 requests per minute. After one minute, you will be able to receive suggestions again.
Note: we are constantly tweaking the performance of the service and monitoring the impact of the rate limit. This limit may change over time.
I use a proxy server and I am having trouble with Ansible Lightspeed; what do I do to connect?¶
The VS Code Ansible extension does not directly implement any networking or communication. If your organization uses a proxy server, then you will need to ensure that your OS settings, browser settings, and VS Code settings properly implement your proxy server. The configuration will differ per-organization, and is something that your local IT department may need to assist with. You may want to consider the VS Code proxy documents for starting VS Code with different flags in order to properly configure networking.
Ansible Lightspeed with IBM watsonx Code Assistant will work behind a proxy server provided that the following domains are allowed through the proxy in order for VS Code to communicate with the service.
What derivatives of VS Code are supported?¶
- VS Code
- VS Code Insiders
- VS Codium
Server-hosted derivatives of VS Code, such as code-server are not supported at this time.
Can I use a GitHub Enterprise account?¶
The Ansible Lightspeed Technical Preview will only work with a public github.com account. GitHub Enterprise authentication is not supported.
What do Red Hat and IBM do with the permissions granted to the GitHub OAuth application used to sign in?¶
Ansible Lightspeed uses GitHub accounts for authorization purposes as the identity provider. GitHub is not used to store or manage any data related to the service.
GitHub requires that an OAuth application request a minimum set of permissions in order to exist as a connected application.
Ansible Lightspeed does not use OAuth permissions to read organizational repository data or use GitHub API data for any other purpose. GitHub is only used for authentication and your GitHub username is used as your username for access to Ansible Lightspeed.
What is the content matching score and how is it calculated?¶
The score returned from a k-nearest neighbor (kNN) search that is generated by a similarity function (which determines the distance between two vectors) and then converted into a search score; larger scores imply a greater similarity between two strings.
This information can be used to infer that a recommendation is likely influenced by a particular training source, but it not an indicator of direct attribution due to the nature of how content is generated by language models.
What protocols and ports does Ansible Lightspeed use?¶
The Ansible extension communicates with Ansible Lightspeed via HTTPS over port 443.
I need more help. Where can I go?¶
The Ansible VS Code extension provides a Contact Us link on the right-hand side of VS Code Status bar. To send an email via the Contact Us button, click on Lightspeed and then on the Contact Us button.
Alternatively, contact us in the Ansible Lightspeed with IBM watsonx Code Assistant Matrix room: