Learning rate on the AI Learning track. The learning rate is the size of the nudge. Too large and loss jumps or diverges. Too small and training crawls and may look stuck. It is the first hyperparameter to get right.
This lesson assumes you already worked through Gradients without the mythology.
The idea in practice
Start with a range used for your optimizer and model size. Plot loss for a few rates on a small run. Pick a rate that decreases smoothly.
A concrete check
goal = {
'track': 'AI Learning',
'lesson': 'Learning rate',
}
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
Copying a learning rate from a paper that used a different batch size and model. 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
Describe the loss curve you expect from a rate that is too high.
Self-check
- Say Learning rate 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.