Training time to a fixed accuracy for perturbative gradient methods grows far slower than linearly with network size, challenging a long-standing scaling objection.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Scaling of hardware-compatible perturbative training algorithms
Training time to a fixed accuracy for perturbative gradient methods grows far slower than linearly with network size, challenging a long-standing scaling objection.