AI Paper Atlas

Research map / Architectures and efficiency / Pruning and efficient tuning

AC/DC: Alternating Compressed/DeCompressed Training of Deep Neural Networks

Alexandra Peste, Eugenia Iofinova, Adrian Vladu, Dan Alistarh

NeurIPS 2021 · Architectures and efficiency › Pruning and efficient tuning · approach: Sparse training from scratch · 10 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
Use alternating dense and sparse training phasesResNet50 · ImageNet, 80% sparsityTop-1 accuracy (%): 74.6 \pm 0.06 -> 76.3 \pm 0.1Improved
Use alternating dense and sparse training phasesResNet50 · ImageNet, 90% sparsityTop-1 accuracy (%): 74.76 -> 75.03 \pm 0.1Improved
Use alternating dense and sparse training phasesResNet50 · ImageNet, 95% sparsityTop-1 accuracy (%): 72.12 -> 73.14 \pm 0.2Improved
Use alternating dense and sparse training phasesResNet50 · ImageNet, 98% sparsityTop-1 accuracy (%): 67.06 -> 68.44 \pm 0.09Improved
Use alternating dense and sparse training phasesMobileNetV1 · ImageNet, 75% sparsityTop-1 accuracy (%): 70.09 -> 70.3 \pm 0.07Improved
Use alternating dense and sparse training phasesMobileNetV1 · ImageNet, 90% sparsityTop-1 accuracy (%): 63.87 -> 66.08 \pm 0.09Improved
Prune embeddings to 50% sparsity (ablation)AC/DC, Transformer-XL · WikiText-103, 80% model sparsityperplexity, sparse: 20.65 -> 20.83Hurt
Prune embeddings to 50% sparsity (ablation)AC/DC, Transformer-XL · WikiText-103, 90% model sparsityperplexity, sparse: 22.32 -> 22.84Hurt
Shorten dense phases from five to two epochs and omit the longer final decompression phase (ablation)AC/DC, ResNet50 · ImageNet, 90% global sparsityTop-1 accuracy (%): 75.03 -> 74.6Hurt
Use uniform rather than global sparsity (ablation)AC/DC, ResNet50 · ImageNet, 90% sparsityTop-1 accuracy (%): 75.03 -> 75.04No clear effect
Keep the first and last layers dense (ablation)AC/DC, ResNet50 · ImageNet, 90% global sparsityTop-1 accuracy (%): 75.03 -> 75.64Improved
Keep the first layer and depth-wise convolutions dense (ablation)AC/DC, MobileNetV1 · ImageNet, 90% sparsityTop-1 accuracy (%): almost 0.5%Improved

Later papers that compared with this method

6 later accepted papers report AC/DC as a baseline in their results tables; 5 of them report a better number on most of the results they share with it.

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.

Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.

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