Learning rate
How far each optimiser step moves the parameters. The single hyperparameter most likely to be the reason a run failed.
What people get wrong
Too high shows up as a loss that spikes or goes to NaN; too low as a loss that falls smoothly but far too slowly. Both look like "training is broken" and need opposite fixes.
A definition is not an understanding. This is taught properly in Tuning, in order of what matters, part of Hyperparameter tuning.
Learn this properlyTerms from the same topic