Narrowing the Focus: Learned Optimizers for Pretrained Models
This work addresses the problem of inefficient optimizer tuning in deep learning for researchers and practitioners, offering a specialized approach that is incremental but improves performance in specific domains.
The paper tackles the challenge of designing effective learned optimizers by specializing them to specific training environments, resulting in a novel optimizer that outperforms traditional methods like Adam and existing general learned optimizers on image classification tasks with robust generalization.
In modern deep learning, the models are learned by applying gradient updates using an optimizer, which transforms the updates based on various statistics. Optimizers are often hand-designed and tuning their hyperparameters is a big part of the training process. Learned optimizers have shown some initial promise, but are generally unsuccessful as a general optimization mechanism applicable to every problem. In this work we explore a different direction: instead of learning general optimizers, we instead specialize them to a specific training environment. We propose a novel optimizer technique that learns a layer-specific linear combination of update directions provided by a set of base optimizers, effectively adapting its strategy to the specific model and dataset. When evaluated on image classification tasks, this specialized optimizer significantly outperforms both traditional off-the-shelf methods such as Adam, as well as existing general learned optimizers. Moreover, it demonstrates robust generalization with respect to model initialization, evaluating on unseen datasets, and training durations beyond its meta-training horizon.