Constraining transformer projection weights to a shared low-rank subspace reportedly enables near-lossless compression of pipeline-parallel communication, matching centralized convergence at 80Mbps bandwidth.
Low-rank gradient descent
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2025 1verdicts
REJECT 1representative citing papers
citing papers explorer
-
Subspace Networks: Scaling Decentralized Training with Communication-Efficient Model Parallelism
Constraining transformer projection weights to a shared low-rank subspace reportedly enables near-lossless compression of pipeline-parallel communication, matching centralized convergence at 80Mbps bandwidth.