first-order-algorithms

This paper analyzes first-order meta-learning algorithms for few-shot learning, introducing Reptile and providing theoretical insights into why these computationally efficient methods work well on established benchmarks.