Visualization is an efficient way of learning about what happens in the machine learning model. The progress of the training process can be tracked in terms of the accuracy or loss value of the target function. Seeing how the elements of a tensor are distributed can also provide us with some insight into how the machine learning algorithm runs. In this section, we are going to look at tfjs-vis, which is a visualization tool that's been designed especially for the TensorFlow.js framework.
As is often the case, tfjs-vis can be installed using npm. It provides UI components that can be easily and seamlessly rendered in our machine learning application. The tool has a pane on the right-hand side of the UI. Here, we can add any number of components to show the metrics of the machine learning model.
First, the layer inspection section provides information about...
