4 ms·
Only because you have chosen for the learning rate to decrease over time so the model converges. You could reset it. I would even argue the LR is not relevant
by vankessel 7y ago
Only because you have chosen for the learning rate to decrease over time so the model converges. You could reset it.
I would even argue the LR is not relevant to plasticity, it's a meta-variable for training the model and not to do with the model itself.
A plastic model would be one that could accomplish a given task after being trained on it, and then trained on a second task while maintaining the ability to accomplish the first task.