Not really. That's called few shot learner.

It's basically unrelated to what happens during training, which is using gradients.