问题:

What is the best way to quickly see the updated graph in the most recent event file in an open TensorBoard session? Re-running my Python app results in a new log file being created with potentially new events/graph. However, TensorBoard does not seem to notice those differences, unless restarted.

回答1:

It turns out that TensorBoard backend refreshes the logs every minute. This has been reported as a TensorFlow issue.

The reload interval can be configured using the --reload_interval flag of the TensorBoard process, but this option is currently only available in master and as of version 0.8 has not been released.

回答2:

My issue is different. Each time I refresh 0.0.0.0:6006, it seems the new graph keep appending to the old one, which is quite annoying.

After trying kill process and delete old log several times, I realized the issue comes from writer.add_graph(sess.graph), because I didn't reset the graph in jupyter notebook. After resetting, the tensorboard could show the newest gragh.