@article{densevssparsepretrainingattinyscaleactiv, title = {Dense vs Sparse Pretraining at Tiny Scale: Active-Parameter vs Total-Parameter Matching}, author = {Abdalrahman Wael}, year = {2026}, eprint = {2605.13769}, archivePrefix = {arXiv}, url = {https://arxiv.org/abs/2605.13769}, }