PowerTrip dynamically selects and scales the set of geo-distributed training sites to minimize time-to-accuracy, beating static baselines by up to 44% in simulation.
QSGD: Communication-Efficient SGD via Gradient Quantization and Encoding
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
citation-role summary
background 1
citation-polarity summary
fields
cs.DC 1years
2025 1verdicts
CONDITIONAL 1roles
background 1polarities
support 1representative citing papers
citing papers explorer
-
PowerTrip: Exploiting Federated Heterogeneous Datacenter Power for Distributed ML Training
PowerTrip dynamically selects and scales the set of geo-distributed training sites to minimize time-to-accuracy, beating static baselines by up to 44% in simulation.