Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Programmatic sampling of model traces #54

Closed
andrey-kuprianov opened this issue Jul 26, 2022 · 0 comments · Fixed by #79
Closed

Programmatic sampling of model traces #54

andrey-kuprianov opened this issue Jul 26, 2022 · 0 comments · Fixed by #79
Assignees

Comments

@andrey-kuprianov
Copy link
Contributor

andrey-kuprianov commented Jul 26, 2022

As documented in ADR-05 Test execution, there is a dependency on the Model module, for providing programmatic access to sampling the traces from the model.

The dependency needs to be implemented via reexporting functionality of Modelator, and combining it with the defaults stored in Atomkraft, in approximately the following form:

get_model_trace(model = None, config = None, samples = None)

where:

  • the first two parameters are file paths, the third one is a list of operator names
  • when not given explicitly, the parameters should be picked up from the current defaults stored in Atomkraft config, where they are set via atomkraft model subcommands.

Return value:

  • when any parameter is not given, and is not available via defaults, or the parameter cannot be parsed, the function should raise an exception explaining the reason of the error.
  • when the model can be sampled for traces given the provided parameters, ModelResult from Modelator API should be returned.
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

Successfully merging a pull request may close this issue.

2 participants