Introduction to Training Job Visualization

ModelArts notebook of the new version supports TensorBoard and MindInsight for visualizing training jobs. In a development environment, use a small dataset to train and debug an algorithm. This is used to check algorithm convergence and detect training issues, facilitating debugging.

ModelArts visualization jobs support TensorBoard and MindInsight.

Both TensorBoard and MindInsight can effectively display the change trend of a training job and the data used in the training.

• TensorBoard

  TensorBoard effectively displays the computational graph of TensorFlow in the running process, the trend of all metrics in time, and the data used in the training. For more details about TensorBoard, see TensorBoard official website.

  TensorBoard visualization training jobs support only CPU and GPU flavors based on TensorFlow 2.1, and PyTorch 1.4 and 1.8 images. Select images and flavors based on the site requirements.

• MindInsight

  MindInsight visualizes information such as scalars, images, computational graphs, and model hyperparameters during training. It also provides functions such as training dashboard, model lineage, data lineage, and performance debugging, helping you train and debug models efficiently. MindInsight supports MindSpore training jobs. For more information about MindInsight, see MindSpore official website.

  The following shows the images and flavors supported by MindInsight visualization training jobs, and select images and flavors based on the site requirements.

  • MindSpore 1.2.0 (CPU or GPU)

You can use the summary file generated during model training to create a visualization job in Notebook of DevEnviron.

• For details about how to create a MindInsight visualization job in a development environment, see MindInsight Visualization Jobs.
• For details about how to create a TensorBoard visualization job in a development environment, see TensorBoard Visualization Jobs.