AI Paper Atlas

Research map / Architectures and efficiency / RNNs and normalization

A Deep Learning Dataloader with Shared Data Preparation

Jian Xie, Jingwei Xu, Guochang Wang, Yuan Yao, Zenan Li, Chun Cao, Hanghang Tong

NeurIPS 2022 · Architectures and efficiency › RNNs and normalization · approach: Efficient network compression and training · 0 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
Replace PyTorch default Baseline dataloader with JOADERResNet18 training · ImageNet, 6 parallel jobsepoch time (min): 67.85 -> 37.55Improved
Replace PyTorch default Baseline dataloader with JOADERResNet18 training · ImageNet, 6 parallel jobsCPU utilization: over 40% less CPU utilizationImproved
JOADER instead of CoorDL when jobs start simultaneouslyResNet18 training · ImageNet, 6 parallel jobsepoch time (min): 37.98 -> 37.55No clear effect
JOADER instead of CoorDL for a single jobResNet18 training · ImageNet, 1 jobepoch time (min): 32.82 -> 32.70No clear effect
JOADER instead of CoorDL with 4 and 5 jobsResNet18 training · ImageNet, 4-5 jobsepoch time (min): 34.76 -> 36.32Hurt
JOADER instead of CoorDL, models with different speedsResNet18 within five-ResNet mix · ImageNet, ResNet18-152speed: 4x fasterImproved
Larger models in the mixJOADER · ImageNet, ResNet101/152benefitHurt
JOADER vs Baseline when jobs arrive at different momentsResNet18 training · ImageNet, four jobstraining time: saves 17.1%Improved
JOADER vs Baseline when jobs arrive at different momentsResNet18 training · ImageNet, four jobsCPU utilization: saves 8.7%Improved
Increase dataset overlap (intersection size) between two jobsJOADER · ImageNet resampledCPU utilization: 60% for same dataset vs different datasetsImproved
Half of datasets in intersection for 2 jobsJOADER vs Baseline · ImageNet resampledCPU utilization: 10% reductionImproved
Remove DSA or RefCnt (ablation) (ablation)JOADER · ImageNettraining efficiencyImproved

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy