Zhongxin Guo

Papers on Lattice

Total citations

Topics

h-index

Research focus

Architecture Design (Transformers, SSMs, MoE) (1)Distributed Systems & Hardware (1)Training Efficiency & Optimization (1)

Frequent co-authors

L. Qu (1)Lianhai Ren (1)Peng Cheng (1)Rui Gao (1)

Papers (1)

Dec 15, 2025

Microsoft ResearchDec 15, 2025·also Deakin

SIGMA: An AI-Empowered Training Stack on Early-Life Hardware

Train massive models on unproven hardware with confidence: SIGMA achieves 94% accelerator utilization and trains a 200B MoE with near-perfect stability on early-life AI accelerators.

L. Qu, Lianhai Ren, Peng Cheng +12

Architecture Design (Transformers, SSMs, MoE)Distributed Systems & Hardware Training Efficiency & Optimization

Search

Zhongxin Guo

Research focus

Frequent co-authors

Papers (1)