AI Paper Atlas

Research map / Architectures and efficiency / RNNs and normalization

Top-KAST: Top-K Always Sparse Training

Siddhant Jayakumar, Razvan Pascanu, Jack Rae, Simon Osindero, Erich Elsen

NeurIPS 2020 · Architectures and efficiency › RNNs and normalization · approach: Efficient network compression and training · 21 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
select extra backward units randomly instead of by magnitude (ablation)Top-KAST · ImageNet ResNet-50, fwd 90% bwd 80%top-1 accuracy: 73.03 -> 74.76Improved
select extra backward units by magnitude (vs random) (ablation)Top-KAST · ImageNet ResNet-50, fwd 95% bwd 90%top-1 accuracy: 68.48 -> 70.42Improved
allow exploration updates to extra units for first 5000 steps (ablation)Top-KAST · ImageNet ResNet-50, fwd 90% bwd 0%top-1 accuracy: 68.26 -> 72.05Improved
allow exploration updates until step 16000 (ablation)Top-KAST · ImageNet ResNet-50, fwd 90% bwd 0%top-1 accuracy: 68.26 -> 74.14Improved
allow exploration for the full 32000 steps vs 16000 (ablation)Top-KAST · ImageNet ResNet-50, fwd 90% bwd 0%top-1 accuracy: 74.14 -> 74.65Improved
compute top-K every 100 steps instead of every step (ablation)Top-KAST · ImageNet ResNet-50, fwd 80% bwd 50%top-1 accuracy: 75.03 -> 75.14No clear effect
compute top-K every 100 steps instead of every step (ablation)Top-KAST · ImageNet ResNet-50, fwd 95% bwd 90%top-1 accuracy: 70.42 -> 70.38No clear effect
use sparse backward pass (bwd 80%) instead of dense backwardTop-KAST · WikiText-103perplexity: 19.8 -> 21.3Hurt
increase forward sparsity to 95% (bwd 90%)Top-KAST · WikiText-103perplexity: 19.8 -> 32.2Hurt
Top-KAST sparse model vs smaller dense modelTop-KAST 80% sparse 57M · WikiText-103perplexity: 21.5 -> 19.8Improved
use Top-KAST with dense backward vs magnitude pruningTop-KAST · enwik8 69M model, fwd 80%BPC: 1.02 -> 1.03Hurt
Top-KAST sparse transformer vs denseTop-KAST (80%, 0%) 55M · enwik8BPC: 0.99 -> 1.00No clear effect

Later papers that compared with this method

4 later accepted papers report Top-KAST as a baseline in their results tables; 4 of them report a better number on most of the results they share with it.

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