yes. this is just raw implementation of the model arch as described in papers. for complete model training with back propogation we need training pipeline with

yes. this is just raw implementation of the model arch as described in papers. for complete model training with back propogation we need training pipeline with optmizer and loss calculation.