Checkpoints on the How to Train Models track. A checkpoint is weights plus the preprocessor plus enough metadata to reload. The best validation checkpoint is the one you ship, not the last epoch.
This lesson assumes you already worked through Regularization during training.
The idea in practice
Save optimizer state only if you plan to resume. Always save the metric and the step.
A concrete check
goal = {
'track': 'How to Train Models',
'lesson': 'Checkpoints',
}
checks = [
'input available at decision time',
'score matches the real decision',
'failure case written down',
]
print(goal['lesson'])
for item in checks:
print('-', item)
Run the sketch locally if you have Python. The printout is a reminder of the checks, not a trained model. Replace the strings with the real inputs from your own example before you treat it as a design.
What usually goes wrong
Overwriting the only checkpoint after a bad epoch. When this happens, stop adding parameters or tools. Fix the check, the data, or the permission, then run the same example again.
What to write down
- The input you are allowed to use at decision time.
- The output and the score or pass rule.
- One failure you will test on purpose.
- What you will not claim the system can do.
Practice
List the files that must exist to reload a model in production.
Self-check
- Say Checkpoints in one sentence that mentions an input and an output.
- Name the failure mode in this lesson and the check that would catch it.
Done when: you can explain this lesson without the page open, and you have a written failure case.