AI Paper Atlas

Research map / Optimization / Optimizers

On the Adequacy of Untuned Warmup for Adaptive Optimization

Jerry Ma, Denis Yarats

AAAI 2021 · Optimization › Optimizers · approach: Adam convergence and adaptive learning rates · 34 citations (OpenAlex)

Read the paper

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
Replace RAdam with untuned exponential or linear warmup in AdamAdam on ResNet-50 · ImageNettop-1 error: 28.2 % ± 0.1; 28.3 % ± 0.1; 28.4 % ± 0.1No clear effect
Replace RAdam with untuned exponential or linear warmup in AdamAdam on Transformer LM · WIKITEXT-103validation perplexity: 20.0 ± 0.0; 20.0 ± 0.0; 20.1 ± 0.1No clear effect
Replace RAdam with untuned exponential or linear warmup in AdamAdam on Transformer · WMT16 EN-DEBLEU: 26.2 ± 0.2; 26.4 ± 0.0; 26.4 ± 0.0No clear effect
Vary the first four RAdam iterations among doing nothing, Adam at the fifth-step rate, or gradually warming up to that rate (ablation)RAdam · Experimental settings described in the paperpractical performanceNo clear effect
Replace the adaptive variance-based account of warmup with analysis of actual Adam update magnitudesAdam warmup analysis · Simulated local minimum and EMNISTupdate magnitude: 1 · α; ≈ 0.16 · α; 0.153 · α; around 0.15 · αMixed

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