The transformer encoder block effectively attains the long-term dependencies from the 1D embedded features with gated linear unit (GLU)-based multi-head self-attention where deep features are learned ...