RNNbow: Visualizing Learning Via Backpropagation Gradients in RNNs

Dylan Cashman, Genevieve Patterson, Abigail Mosca, Nathan Watts, Shannon Robinson, Remco Chang

Abstract: We present RNNbow, an interactive tool for visualizing the gradient flow during backpropagation in training of recurrent neural networks. By visualizing the gradient, as opposed to activations, RNNbow offers insight into how the network is learning. We show how it illustrates the vanishing gradient and the training process.