You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Currently, environments in all of InnerEye and hi-ml do not use environment locking. This means that if a new environment is built (or rebuilt), then secondary dependencies can change and result in inconsistent and even breaking behaviour between 2 deployments of the same code. Locking use poetry and conda-lock isn't an option, as AzureML can only consume files in the format of a conda environment.yml.
Currently, environments in all of InnerEye and hi-ml do not use environment locking. This means that if a new environment is built (or rebuilt), then secondary dependencies can change and result in inconsistent and even breaking behaviour between 2 deployments of the same code. Locking use poetry and conda-lock isn't an option, as AzureML can only consume files in the format of a conda
environment.yml.