AI Paper Atlas

Research map / Architectures and efficiency / RNNs and normalization

DASH: Faster Shampoo via Batched Block Preconditioning and Efficient Inverse-Root Solvers

Ionut-Vlad Modoranu, Philip Zmushko, Erik Schultheis, Mher Safaryan, Dan Alistarh

ICML 2026 · Architectures and efficiency › RNNs and normalization · approach: Scalable approximation and equivariant compute · 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
batch block preconditioners (DASH) instead of sequential Distributed Shampoo, CN FP32Distributed Shampoo · Llama-953M C4, B=1024optimizer step ms: 666 -> 136Improved
DASH with CN FP16 versus DIST CN FP32 at B=1024Distributed Shampoo · Llama-953M C4, B=1024optimizer step ms: 666 -> 119Improved
DASH on EVD f=1, B=2048Distributed Shampoo EVD · Llama-953M C4optimizer step ms: 2200 -> 1747Improved
DASH versus DIST on validation perplexity (EVD, B=2048, f=1)Distributed Shampoo EVD · Llama-953M C4validation perplexity: 11.72 -> 11.73No clear effect
use NDB with Power-Iteration scaling instead of Frobenius scaling (ablation)DASH NDB · Llama-953M C4, B=1024validation perplexity: 11.76 -> 11.68Improved
use NDB instead of CN-FP32 as inverse rootDASH · Llama-953M C4, B=1024validation perplexity: 11.87 -> 11.68Improved
use FP16 instead of FP32 iterations for CN (ablation)DASH CN · Llama-953M C4, B=1024optimizer step ms: 136 -> 119Improved
use BF16 instead of FP16 for CN iterations (ablation)DASH CN · Llama-953M C4convergenceHurt
use FP16 or BF16 for NDB iterations (ablation)DASH NDB · Llama-953M C4convergenceHurt
increase block size from 1024 to 2048 (ablation)DASH · Llama-953M C4optimizer step ms: NDB PIM 177 -> 267Hurt
replace SHMP dampening with RELU heuristic (ablation)DASH EVD · Llama-373Mvalidation perplexity: 15.893 -> 15.876No clear effect
replace Distributed Shampoo with DASH on NDB, ViT-S ImageWoofDistributed Shampoo NDB · ImageWoof ViT-Sopt step ms: 145 -> 15Improved

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