Scholar Feed

Mixture-of-experts routing

GShard

GShard: Scaling Giant Models with Conditional Computation and Automatic Sharding

Heavily superseded#5 of 1,370 most-superseded · first seen Jun 30, 2020

Heavily superseded — a standard baseline that newer methods routinely beat

2 papers critique it · 2 beat it on benchmarks

What papers say

Verbatim critique sentences, each from a paper that cites GShard as a baseline.

this interferes with the model's training objective and degrades accuracy
— GEM: GPU-Variability-Aware Expert to GPU Mapping for MoE Systems
“token dropping occurs when inputs are routed to capacity-saturated experts, while padding operations in underutilized experts create hardware inefficiencies”
— Maximum Score Routing For Mixture-of-Experts

Beaten on benchmarks

Head-to-head results where a newer method reports beating GShard. Values are copied from the source paper's tables — verify against the cited paper.

What to use instead

Recent methods in the same sub-problem, not yet superseded in the knowledge base — arXiv benchmark leaders, not vetted production recommendations.