SPICE: Sparse Parameter Gossip for LLM Pretraining

Pretraining language models across geographically distributed resources requires communication-efficient methods that operate under limited and variable internet bandwidth. Decentralized training replaces global collectives with peer exchanges, but full-model messages make larger peer neighborhoods costly. We introduce SPICE, which combines sparse parameter gossip with error feedback through corrected parameter values that remain directly interpretable as peer neighborhoods change. These sparse exchanges make broader neighborhoods affordable, achieving better training quality with less communication than the evaluated dense decentralized baselines. As the replica count grows, SPICEincreases the number of peers contacted while keeping individual messages sparse, limiting the quality degradation observed with dense gossip. Under bandwidth constraints, SPICE achieves higher training throughput than DDP and dense gossip, while sustaining throughput during transient link slowdowns.

Authors

Publication Details

Journal
arXiv (Cornell University)
Published
2026-11-17
DOI
https://doi.org/10.48550/arxiv.2611.23053
Primary Topic
Stochastic Gradient Optimization Techniques
Type
preprint
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
preprint

SPICE: Sparse Parameter Gossip for LLM Pretraining

Eugene Belilovsky, Edouard Oyallon, Pietro Cagnasso, Daniele Kopyshevskiy
arXiv (Cornell University)
Stochastic Gradient Optimization Techniques
preprint

SPICE: Sparse Parameter Gossip for LLM Pretraining

Eugene Belilovsky, Edouard Oyallon, Pietro Cagnasso, Daniele Kopyshevskiy
preprint en

Abstract

Pretraining language models across geographically distributed resources requires communication-efficient methods that operate under limited and variable internet bandwidth. Decentralized training replaces global collectives with peer exchanges, but full-model messages make larger peer neighborhoods costly. We introduce SPICE, which combines sparse parameter gossip with error feedback through corrected parameter values that remain directly interpretable as peer neighborhoods change. These sparse exchanges make broader neighborhoods affordable, achieving better training quality with less communication than the evaluated dense decentralized baselines. As the replica count grows, SPICEincreases the number of peers contacted while keeping individual messages sparse, limiting the quality degradation observed with dense gossip. Under bandwidth constraints, SPICE achieves higher training throughput than DDP and dense gossip, while sustaining throughput during transient link slowdowns.

arXiv (Cornell University)
Stochastic Gradient Optimization Techniques
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon