Make the output of recorder.plot_loss() look also smooth for the training loss

Is there a way to make the plot for the training loss look smoother like the validation loss one. One can hardly see the validation loss when the training loss fluctuates