Tag
This paper investigates the effects of cross-depth aggregation and soft-routed experts on EEG foundation model fine-tuning across multiple datasets, showing inconsistent benefits and significant computational overhead.