← All works
Research paper

Model-Agnostic Meta-Learning for Fast Adaptation of Deep Networks

At a glance

MAML trains models to adapt to new tasks with only a few examples.

In one paragraph

Introduced MAML, training models across different tasks so they can adapt to a new task with only a few examples and a small amount of training.

Authors 3

  1. Chelsea Finn
  2. Pieter Abbeel
  3. Sergey Levine