aidkit-client 0.2.0

aidkit for your CI/CD and j-notebooks.

aidkit is an MLOps platform that allows you to assess and defend against threads and vulnerabilities of AI models before they deploy to production. aidkit-client is a companion python client library to seamlessly integrate with aidkit in python projects.

Changelog

Breaking changes are written in bold text.

Version 0.2.0

• Rename resources.Report.table to resources.Report.table_string.
• Add resources.Report.table function returning a pandas dataframe.
• Move endpoints module to _endpoints.
• Remove class aidkit.Aidkit.
• Add docstrings.
• Rename exceptions.AidkitCLIError to exceptions.AidkitClientError.
• Add class exceptions.AuthenticationError, which is raised whenever the client fails to authenticate to the server.
• Include http error code and http body returned by the aidkit server in all exceptions.
• Make http request timeout configurable.
• Add id and name properties to resources.
• Add option to use progress bar in PipelineRun.report and MLModelVersion.upload.
• Add PipelineRun.get_progress method, which can be used to check how far a pipeline run has progressed.
• Update web API version.
• Add functionality to manage datasets and subsets from the client including data upload with added resources Dataset and Observation and updated resource Subset.
• Add functionality to download artifacts from the report.