SaMoE: Parameter Efficient MoE Language Models via Self-Adaptive Expert Combination
Keywords: Mixture-of-Expert, Autoregressive language model, Parameter efficiency.
Abstract: Recently, Mixture-of-Experts (MoE) has demonstrated success in scaling models to have large amounts of parameters without significant increases in computational cost. However, MoEs have been also reported to be parameter inefficient such that larger models do not always lead to better performance.
In this work, we study how to build parameter-efficient MoE models. Our analysis identifies that MoE layers exhibit poor gradient flow as the number of experts increases, leading to insufficient training of experts. To overcome this issue, we propose a new MoE architecture design (SaMoE), which improves the parameter efficiency of MoE models by learning a soft combination of a global set of expert layers for each MoE layer. Such a scheme enables substantial parameter savings on MoE while achieving comparable or better accuracy than the standard MoE training baseline. Extensive experiments on billion-scale GPT-3 style autoregressive MoE language models demonstrate that SaMoE significantly improves the parameter efficiency of MoE models by reducing up to 5.2X total parameters while obtaining superior pre-training and zero-shot generalization results as compared to baseline.
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.
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
Submission Guidelines: Yes
Please Choose The Closest Area That Your Submission Falls Into: Deep Learning and representational learning
TL;DR: SaMoE is a parameter efficient MoE architecture design that enables parameter savings on MoE while achieving comparable or better accuracy.
16 Replies
Loading