AI Paper Atlas

Research map / Optimization / Implicit bias and generalization

Accumulated Decoupled Learning with Gradient Staleness Mitigation for Convolutional Neural Networks

Huiping Zhuang, Zhenyu Weng, Fulin Luo, Toh Kar-Ann, Haizhou Li, Zhiping Lin

ICML 2021 · Optimization › Implicit bias and generalization · approach: Edge-of-stability dynamics · 1 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
add module-wise gradient accumulation (mGA step M=4 vs M=1-3) (ablation)ADL training RN-56 (K=16) · CIFAR-10validation error (%)Improved
increase mGA step from 4 to 12 (ablation)ADL training RN-56 (K=16) · CIFAR-10validation error (%): 6.38 -> 6.86Hurt
adopt ADL (mGA + asynchronous pipelining) instead of DDGdecoupled learning on RN-56 K=8 · CIFAR-10validation error (%): diverged -> 6.18Improved
adopt ADL instead of FR at large partition sizeRN-56 K=16 · CIFAR-10validation error (%): 11.51 -> 6.38Improved
adopt ADL instead of DGL at large partition sizeRN-56 K=8 · CIFAR-10validation error (%): 13.26 -> 6.18Improved
ADL vs BP on ImageNet at large partition sizeRN-18 K=10 (max) · ImageNetTop1 error (%): 29.79 -> 29.84No clear effect
ADL vs BP on ImageNet RN-50 K=4RN-50 · ImageNetTop1 error (%): 23.65 -> 23.43 (Top1), 7.13 -> 7.45 (Top5)Mixed
adopt ADL (3 lockings removed) instead of DDG/FR for speedRN-101 ImageNet K=2 · ImageNetspeedup over BP: 1.19-1.32x -> 1.92xImproved
ADL vs DGL speed at K=4RN-101 ImageNet · ImageNetspeedup over BP: 3.32x vs 3.39xNo clear effect
adopt ADL instead of BP (memory)RN-18 K=4 · CIFAR/ImageNet memoryGPU memory (GB): 1.14 -> 1.84Hurt

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