On Layer Normalization in the Transformer Architecture

by   Ruibin Xiong, et al.

The Transformer is widely used in natural language processing tasks. To train a Transformer however, one usually needs a carefully designed learning rate warm-up stage, which is shown to be crucial to the final performance but will slow down the optimization and bring more hyper-parameter tunings. In this paper, we first study theoretically why the learning rate warm-up stage is essential and show that the location of layer normalization matters. Specifically, we prove with mean field theory that at initialization, for the original-designed Post-LN Transformer, which places the layer normalization between the residual blocks, the expected gradients of the parameters near the output layer are large. Therefore, using a large learning rate on those gradients makes the training unstable. The warm-up stage is practically helpful for avoiding this problem. On the other hand, our theory also shows that if the layer normalization is put inside the residual blocks (recently proposed as Pre-LN Transformer), the gradients are well-behaved at initialization. This motivates us to remove the warm-up stage for the training of Pre-LN Transformers. We show in our experiments that Pre-LN Transformers without the warm-up stage can reach comparable results with baselines while requiring significantly less training time and hyper-parameter tuning on a wide range of applications.


page 1

page 2

page 3

page 4


Understanding the Difficulty of Training Transformers

Transformers have been proved effective for many deep learning tasks. Tr...

RealFormer: Transformer Likes Residual Attention

Transformer is the backbone of modern NLP models. In this paper, we prop...

On Layer Normalizations and Residual Connections in Transformers

In the perspective of a layer normalization (LN) position, the architect...

Why Deep Transformers are Difficult to Converge? From Computation Order to Lipschitz Restricted Parameter Initialization

The Transformer translation model employs residual connection and layer ...

Population Gradients improve performance across data-sets and architectures in object classification

The most successful methods such as ReLU transfer functions, batch norma...

How to Initialize your Network? Robust Initialization for WeightNorm & ResNets

Residual networks (ResNet) and weight normalization play an important ro...

Guidelines for the Regularization of Gammas in Batch Normalization for Deep Residual Networks

L2 regularization for weights in neural networks is widely used as a sta...

Code Repositories



view repo