Beginning Anomaly Detection Using

embeddings_layer_names: The list of names of layers for 

TensorBoard to track. If None or an empty list, then all of the layers 

will be watched. Set to None by default.

• 

embeddings_metadata: A dictionary that maps layer names to the 

corresponding file names where the metadata for this embedding 

layer is saved. Set to None by default. If the same metadata file is used 

for all of the embedding layers, then a string can be passed.

• 

embeddings_data: The data to be embedded at the layers specified 

in embeddings_layer_names. This is a Numpy array if the model 

expects a single input, and multiple Numpy arrays if the model has 

multiple inputs. Set to None by default.

• 

update_freq: A ‘batch’, ‘epoch’, or integer. ‘batch’ writes the losses and 

metrics to TensorBoard after each batch. ‘epoch’ is the same, except 

the losses and metrics are written to TensorBoard after each epoch. 

The integer tells it to write the metrics and losses to TensorBoard 

every integer