Investigation into the Training Dynamics of Learned Optimizers
This work addresses stability and generalization issues in learned optimizers, which are incremental improvements for deep learning practitioners.
The paper tackled the problem of understanding and improving learned optimizers by analyzing their training dynamics, focusing on network symmetries and update distributions, and identified insights for mutual benefits between learned and hand-designed optimizers.
Optimization is an integral part of modern deep learning. Recently, the concept of learned optimizers has emerged as a way to accelerate this optimization process by replacing traditional, hand-crafted algorithms with meta-learned functions. Despite the initial promising results of these methods, issues with stability and generalization still remain, limiting their practical use. Moreover, their inner workings and behavior under different conditions are not yet fully understood, making it difficult to come up with improvements. For this reason, our work examines their optimization trajectories from the perspective of network architecture symmetries and parameter update distributions. Furthermore, by contrasting the learned optimizers with their manually designed counterparts, we identify several key insights that demonstrate how each approach can benefit from the strengths of the other.