Considerations

Although TensorBoard is an extremely rich and well featured tool for visualizing data and models, it has its limitations that users need to factor in before standardizing on it. Let us review some of them below.

Data and Model Versioning

When tuning a model or setting values for hyperparameters, users may want to save different model and training data versions. Especially when conducting experiments, they may want to look at different versions of the model and data at the same time. When using TensorBoard, users cannot tag a certain run or a set of data as important to be highlighted etc

Unstructured Data Formats

Some data types such as video data cannot be visualized in TensorBoard.

---