AI Paper Atlas

Research map / Optimization / Optimizers

Softsign: Smooth Sign in Your Optimizer For Better Parameter Heterogeneity Handling

Dmitrii Feoktistov, Timofey Belinsky, Andrey Veprikov, Amir Zainullin, Aleksandr Beznosikov

ICML 2026 · Optimization › Optimizers · approach: Second-order and structured preconditioning · 0 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
replace hard sign with smooth soft sign and temperature scheduleSignum · Next-character prediction (Transformer)accuracy (%): 56.17 -> 57.05Improved
replace hard sign with smooth soft sign and temperature scheduleSignum · Next-character prediction (LSTM)accuracy (%): 59.71 -> 61.36Improved
replace hard switch to SGD with smooth transition (ablation)SoftSignum vs HardSwitchSign · Next-character prediction (LSTM)accuracy (%): 60.40 -> 61.36Improved
apply soft spectral transition to MuonMuon · Next-character prediction (Transformer)accuracy (%): 57.36 -> 57.84Improved
apply soft spectral transition to MuonMuon · Next-character prediction (LSTM)accuracy (%): 63.02 -> 63.02No clear effect
apply soft spectral transition to MuonMuon · GraphLand (Graph Transformer)average rank: 2.43 -> 1.86Improved
replace hard sign with soft signSignum · GraphLand (Graph Transformer)average rank: 3.86 -> 2.71Improved
apply soft spectral transition to MuonMuon · LLM pretraining (720M, FineWeb)perplexity: 16.362 -> 16.216Improved
decrease alpha_sign from 0.9 to 0.3 (ablation)SoftSignum · Next-character prediction (Transformer)val accuracy (%): 57.05 -> 57.20No clear effect
decrease alpha_sign from 0.9 to 0.3 (ablation)SoftSignum · LLM pretraining (130M LLaMA, C4)perplexity: 18.519 -> 18.559No clear effect
use SoftSignum instead of SGD under heavy class imbalance (k >= 10)SoftSignum · Unbalanced CIFAR-10F1-scoreImproved
use pure Signum instead of SoftSignum under extreme imbalance (k >= 100)Signum · Unbalanced CIFAR-10F1-scoreHurt

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