Simplifying Transformer Blocks
Code Of Ethics: I acknowledge that I and all co-authors of this work have read and commit to adhering to the ICLR Code of Ethics.
Keywords: transformers, signal propagation theory, self-attention, initialisation, simpler architectures, skip connections, normalisation, fast training speed
Submission Guidelines: I certify that this submission complies with the submission instructions as described on https://iclr.cc/Conferences/2024/AuthorGuide.
TL;DR: We use signal propagation and empirical insights to remove various components of the transformer block without loss of training speed, including: skip connections, values and projections parameters.
Abstract: A simple design recipe for deep Transformers is to compose identical building blocks. But standard transformer blocks are far from simple, interweaving attention and MLP sub-blocks with skip connections \& normalisation layers in precise arrangements. This complexity leads to brittle architectures, where seemingly minor changes can significantly reduce training speed, or render models untrainable.
In this work, we ask to what extent the standard transformer block can be simplified? Combining signal propagation theory and empirical observations, we motivate modifications that allow many block components to be removed with no loss of training speed, including skip connections, projection or value parameters, sequential sub-blocks and normalisation layers. In experiments on both autoregressive decoder-only and BERT encoder-only models, our simplified transformers match the per-iteration training speed and performance of standard transformers, while enjoying 16\% faster training throughput, and using 15\% fewer parameters.
Anonymous Url: I certify that there is no URL (e.g., github page) that could be used to find authors' identity.
No Acknowledgement Section: I certify that there is no acknowledgement section in this submission for double blind review.
Primary Area: unsupervised, self-supervised, semi-supervised, and supervised representation learning
Submission Number: 1634
Loading