English

Layer-wise Quantization for Quantized Optimistic Dual Averaging

Machine Learning 2025-05-21 v1 Optimization and Control

Abstract

Modern deep neural networks exhibit heterogeneity across numerous layers of various types such as residuals, multi-head attention, etc., due to varying structures (dimensions, activation functions, etc.), distinct representation characteristics, which impact predictions. We develop a general layer-wise quantization framework with tight variance and code-length bounds, adapting to the heterogeneities over the course of training. We then apply a new layer-wise quantization technique within distributed variational inequalities (VIs), proposing a novel Quantized Optimistic Dual Averaging (QODA) algorithm with adaptive learning rates, which achieves competitive convergence rates for monotone VIs. We empirically show that QODA achieves up to a 150%150\% speedup over the baselines in end-to-end training time for training Wasserstein GAN on 12+12+ GPUs.

Keywords

Cite

@article{arxiv.2505.14371,
  title  = {Layer-wise Quantization for Quantized Optimistic Dual Averaging},
  author = {Anh Duc Nguyen and Ilia Markov and Frank Zhengqing Wu and Ali Ramezani-Kebrya and Kimon Antonakopoulos and Dan Alistarh and Volkan Cevher},
  journal= {arXiv preprint arXiv:2505.14371},
  year   = {2025}
}

Comments

Accepted at the International Conference on Machine Learning (ICML 2025)

R2 v1 2026-07-01T02:25:08.570Z