AI Paper Atlas

Research map / Optimization / Optimizers

Sketchy: Memory-efficient Adaptive Regularization with Frequent Directions

Vladimir Feinberg, Xinyi Chen, Y. Jennifer Sun, Rohan Anil, Elad Hazan

NeurIPS 2023 · Optimization › Optimizers · approach: Second-order and structured preconditioning · 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
use S-Shampoo (rank 256 FD) instead of AdamAdam · ImageNet v1 validationtop-1 accuracy: 76.76% -> 77.18%Improved
full Shampoo instead of S-ShampooS-Shampoo · ImageNet v1 validationtop-1 accuracy: 77.18% -> 77.69%Improved
S-Shampoo instead of ShampooShampoo · ogbg-molpcbaoverfitting / test APImproved
S-Shampoo instead of AdamAdam · ImageNet v2, Librispeech, ogbg-molpcbatest metricNo clear effect
FD rank 256 for covariance blocks of size 1024Shampoo · ResNet-50, Conformer, GNNmemory: 4x memory savingsImproved
S-Adagrad instead of AdagradAdagrad · gisetteaverage cumulative loss: 0.209 -> 0.158Improved
S-Adagrad instead of AdagradAdagrad · a9aaverage cumulative loss: 0.332 -> 0.333No clear effect
S-Adagrad instead of AdagradAdagrad · cifar10average cumulative loss: 0.303 -> 0.297Improved
S-Adagrad instead of Ada-FDAda-FD · cifar10average cumulative loss: 2.999 -> 0.297Improved
S-Adagrad instead of FD-SONFD-SON · a9aaverage cumulative loss: 0.539 -> 0.333Improved
S-Adagrad instead of RFD-SONRFD-SON · gisetteaverage cumulative loss: 0.167 -> 0.158Improved
S-Adagrad instead of RFD-SONRFD-SON · cifar10average cumulative loss: 0.297 -> 0.297No clear effect

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 0 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