Revisiting the Initial Steps in Adaptive Gradient Descent Optimization
This work addresses instability and generalization issues in adaptive gradient optimization for deep learning practitioners, offering a simple incremental improvement.
The paper tackles the suboptimal generalization and instability of adaptive gradient optimizers like Adam by identifying the standard zero initialization of the second-order moment estimation as a key factor, and it shows that using non-zero initialization strategies improves stability and final performance, with Adam achieving results comparable to recent variants.
Adaptive gradient optimization methods, such as Adam, are prevalent in training deep neural networks across diverse machine learning tasks due to their ability to achieve faster convergence. However, these methods often suffer from suboptimal generalization compared to stochastic gradient descent (SGD) and exhibit instability, particularly when training Transformer models. In this work, we show the standard initialization of the second-order moment estimation ($v_0 =0$) as a significant factor contributing to these limitations. We introduce simple yet effective solutions: initializing the second-order moment estimation with non-zero values, using either data-driven or random initialization strategies. Empirical evaluations demonstrate that our approach not only stabilizes convergence but also enhances the final performance of adaptive gradient optimizers. Furthermore, by adopting the proposed initialization strategies, Adam achieves performance comparable to many recently proposed variants of adaptive gradient optimization methods. Our code is available at https://github.com/Walleclipse/Adam_Initialization.