Gradients without the mythology on the AI Learning track. A gradient says which way to nudge each parameter to reduce loss a little. Training repeats that nudge. You do not need to derive it by hand to use it, but you must know that a zero gradient means this step learned nothing.
This lesson assumes you already worked through Representations.
The idea in practice
Watch gradient norms. If they vanish, the update is tiny. If they explode, the update destroys the weights. Clipping and a smaller step size are the first checks.
A concrete check
goal = {
'track': 'AI Learning',
'lesson': 'Gradients without the mythology',
}
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
Ignoring NaN loss and hoping the next epoch heals it. 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 you would check first if loss becomes NaN on step 20.
Self-check
- Say Gradients without the mythology 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.