Getting the most out of TensorBoard

Hi, I’m currently using TensorBoard to analyse experiments involving the training and evaluation of time series forecasting models. I currently create plots of various metrics in TensorBoard (MSE, MAE etc) but am interested in improving the analysis I do through TensorBoard.

Does anyone have any recommendations of:

  • Useful plugins to add
  • Extra logging to add
  • Ways of customising the TensorBoard plots
  • Anything else that has helped you get the most out of TensorBoard

Thanks in advance!

Hi, have you visited TensorBoard’s official Get started page Get started with TensorBoard  |  TensorFlow ? On the left-hand side - and you may have to expand the browser window to see it - there is a list of different guides. Let us know what you think. In addition, there’s a long tutorial video made by a YouTuber about TensorBoard: TensorFlow Tutorial 17 - Complete TensorBoard Guide - YouTube you may want to check out.

Thank you for your answer, that video was very helpful. I’ve read through the get started guides for TensorBoard, but was more wondering if the community had any more advanced tips.

Do you know if there is any documentation for the Time Series plugin that was added to TensorBoard? It sounds pretty relevant to what I’m working on