A Neural ODE Interpretation of Transformer Layers
This work addresses architectural inefficiencies in transformers for machine learning practitioners, offering an incremental improvement over standard designs.
The paper tackles the problem of improving transformer layer architecture by proposing a parallel arrangement of attention and MLP sublayers, inspired by a Neural ODE interpretation, and shows that this modification enhances performance across multiple tasks, with further gains in image classification using advanced ODE solvers.
Transformer layers, which use an alternating pattern of multi-head attention and multi-layer perceptron (MLP) layers, provide an effective tool for a variety of machine learning problems. As the transformer layers use residual connections to avoid the problem of vanishing gradients, they can be viewed as the numerical integration of a differential equation. In this extended abstract, we build upon this connection and propose a modification of the internal architecture of a transformer layer. The proposed model places the multi-head attention sublayer and the MLP sublayer parallel to each other. Our experiments show that this simple modification improves the performance of transformer networks in multiple tasks. Moreover, for the image classification task, we show that using neural ODE solvers with a sophisticated integration scheme further improves performance.