Investigating Low-Rank Training in Transformer Language Models: Efficiency and Scaling Analysis
This addresses efficiency and scalability issues for researchers and practitioners using Transformer-based LLMs, though it is incremental as it builds on existing low-rank methods in a new context.
The study tackled the high computational costs of large language models by applying low-rank parametrization to feedforward networks in Transformers, achieving a 2.6× speed-up with 32% parameters and showing steeper scaling curves that led to networks surpassing medium and large Transformers in perplexity and throughput.
State-of-the-art LLMs often rely on scale with high computational costs, which has sparked a research agenda to reduce parameter counts and costs without significantly impacting performance. Our study focuses on Transformer-based LLMs, specifically applying low-rank parametrization to the computationally intensive feedforward networks (FFNs), which are less studied than attention blocks. In contrast to previous works, (i) we explore low-rank parametrization at scale, up to 1.3B parameters; (ii) within Transformer language models rather than convolutional architectures; and (iii) starting from training from scratch. Experiments on the large RefinedWeb dataset show that low-rank parametrization is both efficient (e.g., 2.6$\times$ FFN speed-up with 32\% parameters) and effective during training. Interestingly, these structured FFNs exhibit steeper scaling curves than the original models. Motivated by this finding, we develop the wide and structured networks surpassing the current medium-sized and large-sized Transformer in perplexity and throughput performance. Our code is available at https://github.com/CLAIRE-Labo/StructuredFFN/tree/main.