Gradient Sparsification For Masked Fine-Tuning of Transformers | IEEE Conference Publication | IEEE Xplore