AI Paper Atlas

Research map / Architectures and efficiency / Pruning and efficient tuning

Sanity-Checking Pruning Methods: Random Tickets can Win the Jackpot

Jingtong Su, Yihang Chen, Tianle Cai, Tianhao Wu, Ruiqi Gao, Liwei Wang, Jason Lee

NeurIPS 2020 · Architectures and efficiency › Pruning and efficient tuning · approach: Pruning at initialization · 13 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
randomly rearrange kept connections per layer, keeping per-layer counts (random tickets from smart ratios) (ablation)initial tickets (SNIP/GraSP/LT) · CIFAR-10 / CIFAR-100test accuracyNo clear effect
corrupt pruning data with random labels and random pixels (ablation)initial tickets (SNIP/GraSP/LT) · CIFAR-10test accuracyNo clear effect
use data-independent random tickets with smart ratios instead of GraSPVGG19 · CIFAR-10 98% prunedtest accuracy: 91.90 -> 92.45Improved
use random tickets with smart ratios instead of SNIPResNet32 · Tiny-ImageNet 98% prunedtest accuracy: 32.09 -> 43.93Improved
use random tickets instead of GraSPResNet32 · Tiny-ImageNet 98% prunedtest accuracy: 39.43 -> 43.93Improved
use random tickets instead of GraSPResNet32 · CIFAR-10 98% prunedtest accuracy: 89.21 -> 89.10No clear effect
replace descending smart ratio with balanced keep-ratio (ablation)random tickets · CIFAR-10 ResNet32 98%test accuracy: 89.10 -> 86.52Hurt
replace descending smart ratio with ascending keep-ratio (ablation)random tickets · CIFAR-10 ResNet32test accuracy: 89.10 -> 10.00Hurt
use cubic or linear decay instead of smart ratio (ablation)random tickets · CIFAR-10test accuracy: 89.10 -> 89.14No clear effect
shuffle unmasked weights per layer (ablation)learning rate rewinding · CIFAR-10 ResNet32 95%test accuracy: 93.02 -> 91.29Hurt
prune with half the dataset (ablation)learning rate rewinding · CIFAR-10 ResNet32 95%test accuracy: 93.02 -> 92.03Hurt
apply smart-ratio layerwise magnitude pruning (hybrid tickets) instead of globallearning rate rewinding · CIFAR-100 ResNet32 98%test accuracy: 59.22 -> 63.44Improved

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