A unified convergence analysis of LoRA aggregation in federated learning shows Product-Sum aggregation converges globally at the optimal rate, while Sum-Product aggregation suffers from broadcast error from SVD truncation.
Heterogeneous loRA for federated fine-tuning of on-device foundation models
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.LG 1years
2025 1verdicts
CONDITIONAL 1roles
background 1polarities
unclear 1representative citing papers
citing papers explorer
-
Convergence Analysis of Aggregation-Broadcast in LoRA-enabled Distributed Fine-Tuning
A unified convergence analysis of LoRA aggregation in federated learning shows Product-Sum aggregation converges globally at the optimal rate, while Sum-Product aggregation suffers from broadcast error from SVD truncation.