GamMa: Efficient Fine-Tuning of Pre-Trained Language Models Using Gradient Activation Mapping Masking | IEEE Conference Publication | IEEE Xplore