LoRA dropout as a sparsity regularizer for overfitting reduction

Published: 01 Jan 2025, Last Modified: 11 Oct 2025Knowl. Based Syst. 2025EveryoneRevisionsBibTeXCC BY-SA 4.0
Abstract: Highlights•Generalization error bound of applying dropout on LoRA.•A more expressive and efficient design to implement dropout on LoRA.•Extensive experiments to verify the effectiveness of LoRA Dropout.
Loading