Brain network science modelling of sparse neural networks enables Transformers and LLMs to perform as fully connected

This study aims to enlarge our current knowledge on the application of braininspired network science principles for training artificial neural networks (ANNs) with sparse connectivity.