ASI compresses training activations with a single warm-started subspace iteration and a once-per-model rank selection, cutting on-device training memory by up to 120x and FLOPs by up to 1.86x on standard benchmarks.
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
-
Beyond Low-rank Decomposition: A Shortcut Approach for Efficient On-Device Learning
ASI compresses training activations with a single warm-started subspace iteration and a once-per-model rank selection, cutting on-device training memory by up to 120x and FLOPs by up to 1.86x on standard benchmarks.