Artificial Intelligence By Example

TensorBoard Projector

TensorBoard projector loads the metadata from the output of the Embedding.py located in the log directory. First, run Embedding.py. Then run Tensorboard.py. For this example, click on Load data and load the labels.tsv file from the log directory generated by Embedding.py.

Next, TensorBoard runs a PCA calculation (the option on the left in the TensorBoard window) as explained previously. It then displays a 3D view, as shown in the following screenshot:

PCA 3D view

You can now select words on the right-hand side of the window, move the projection around in 3D, zoom, and explore the workplace mind of X.