paper-with-me

Papers

Accelerating Large Batch Training via Gradient Signal to Noise Ratio (GSNR)

2023-09-24 · Guo-qing Jiang, Jinlong Liu, Zixiang Ding, Lin Guo, Wei Lin

As models for nature language processing (NLP), computer vision (CV) and recommendation systems (RS) require surging computation, a large number of GPUs/TPUs are paralleled as a large batch (LB) to improve training throughput. However, training such LB tasks often meets large generalization gap and downgrades final precision, which limits enlarging the batch size. In this work, we develop the variance reduced gradient descent technique (VRGD) based on the gradient signal to noise ratio (GSNR) and apply it onto popular optimizers such as SGD/Adam/LARS/LAMB. We carry out a theoretical analysis of convergence rate to explain its fast training dynamics, and a generalization analysis to demonstrate its smaller generalization gap on LB training. Comprehensive experiments demonstrate that VRGD can accelerate training ($1\sim 2 \times$), narrow generalization gap and improve final accuracy. We push the batch size limit of BERT pretraining up to 128k/64k and DLRM to 512k without noticeable accuracy loss. We improve ImageNet Top-1 accuracy at 96k by $0.52pp$ than LARS. The generalization gap of BERT and ImageNet training is significantly reduce by over $65\%$.

📄 PDF Abstract BibTeX arXiv:2309.13681

Code (0)

등록된 구현이 없습니다.

Tasks

Recommendation Systems

Methods 이 논문이 사용한 방법론

Multi-Head Attention 설명 없음
Attention 설명 없음
Layer Normalization Unlike batch normalization, Layer Normalization directly estimates the normalization statistics from the summed inputs…
WordPiece 설명 없음
Dense Connections Dense Connections, or Fully Connected Connections, are a type of layer in a deep neural network that use a linear operation where every input is connected to every output…
Linear Layer A Linear Layer is a projection $\mathbf{XW + b}$.
Softmax The Softmax output function transforms a previous layer's output into a vector of probabilities. It is commonly used for multiclass classification. Given an input vector $x$…
Residual Connection 설명 없음

Similar Papers 제목 키워드 기반

DIVEBATCH: Accelerating Model Training Through Gradient-Diversity Aware Batch Size Adaptation

2025-09-19 · Yuen Chen, Yian Wang, Hari Sundaram arxiv

The goal of this paper is to accelerate the training of machine learning models, a critical challenge since the training of large-scale deep neural models can be computationally expensive. Stochastic gradient descent (SG…

Computational Efficiency

Accelerating Minibatch Stochastic Gradient Descent using Typicality Sampling

2019-03-11 · Xinyu Peng, Li Li, Fei-Yue Wang

Machine learning, especially deep neural networks, has been rapidly developed in fields including computer vision, speech recognition and reinforcement learning. Although Mini-batch SGD is one of the most popular stochas…

Reinforcement LearningReinforcement Learning (RL)speech-recognitionSpeech Recognition+1

Dynamically Adjusting Transformer Batch Size by Monitoring Gradient Direction Change

2020-05-05 · ACL 2020 6 · Hongfei Xu, Josef van Genabith, Deyi Xiong, Qiuhui Liu

The choice of hyper-parameters affects the performance of neural models. While much previous research (Sutskever et al., 2013; Duchi et al., 2011; Kingma and Ba, 2015) focuses on accelerating convergence and reducing the…

Never Skip a Batch: Continuous Training of Temporal GNNs via Adaptive Pseudo-Supervision

2025-05-18 · Alexander Panyshev, Dmitry Vinichenko, Oleg Travkin, Roman Alferov 외

Temporal Graph Networks (TGNs), while being accurate, face significant training inefficiencies due to irregular supervision signals in dynamic graphs, which induce sparse gradient updates. We first theoretically establis…

Graph Learning

CATERPILLAR: Coarse Grain Reconfigurable Architecture for Accelerating the Training of Deep Neural Networks

2017-06-01 · Yuan-Fang Li, Ardavan Pedram

Accelerating the inference of a trained DNN is a well studied subject. In this paper we switch the focus to the training of DNNs. The training phase is compute intensive, demands complicated data communication, and conta…