I need to run my machine learning program in tensorflow and write the summary in a file (to display for tensorboard). Assume that the program can stop freely at any time.
However, the problem is Tensorflow will write a new summary in a as a new file. Tensorboard will show only linegraph the latest summary (overwriting the previous ones). I want to show both the latest summary and the previous one.
I would be grateful if anyone can provide some guidance on this problem
Save the summaries to the different subdirectories, e.g. for the first run, save to logs/run1
, and for the second to logs/run2
. When you start tensorboard as:
tensorboard --logdir logs
Then both of your runs will be depicted on common plots (where applicable).