What an optimizer does on the How to Train Models track. An optimizer applies gradients to parameters. SGD is the plain step. Adam adapts the step per parameter. Neither fixes bad data.
This lesson assumes you already worked through Pick a loss you can explain.
The idea in practice
Start with the optimizer your architecture tutorial used, then change one thing at a time.
A concrete check
goal = {
'track': 'How to Train Models',
'lesson': 'What an optimizer does',
}
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
Blaming the optimizer for a leaked feature. 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
Say what is updated on each step, and what is not.
Self-check
- Say What an optimizer does 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.