Skip to content

fix: preserve checkpoint_config when HyperparameterTuner defines training jobs - #6389

Open
toddstep wants to merge 2 commits into
aws:masterfrom
toddstep:fix_tuner_checkpoint
Open

toddstep wants to merge 2 commits into
aws:masterfrom
toddstep:fix_tuner_checkpoint

Conversation

@toddstep

@toddstep toddstep commented Oct 7, 2026

Copy link
Copy Markdown
Contributor

HyperparameterTuner._build_training_job_definition() is not using the checkpoint_config attribute of model_trainer.

This causes training jobs to run without requested checkpoints.

Issue #, if available:
N/A

Description of changes:
Include the checkpoint_config parameter in the training job definition
Add a unit test

By submitting this pull request, I confirm that you can use, modify, copy, and redistribute this contribution, under the terms of your choice.

The HyperparameterTuner class copies attributes from its model_trainer
argument. This unit test checks whether checkpoint_config, if provided,
is included in the training job definition.
The HyperparameterTuner class copies attributes from its model_trainer
argument. If checkpoint_config is defined in model_trainer, then
it needs to be explicitly provided in the training job definition.

This branch is waiting to be deployed

1 waiting deployment
manual-approval — 09e5959f Waiting Oct 7, 2026 by toddstep via wait-for-approval #1928
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 this pull request may close these issues.

1 participant