Most ML graph visualizations are all wrong. No one is doing it well. The point of visualizing the graph is to see the data flows. Good luck tracing that in most of the graphviz-style outputs you currently get. How to do it: Show an image, flowing through each layer, plus shape.
-
Show this thread
-
Show of hands: "I have used the 'graph' tab on Tensorboard, and it helped me understand my model better."
2 replies 1 retweet 1 likeShow this thread
Replying to @theshawwn
For understand a program which I have written in the TensorFlow GraphDef language, the TensorBoard graph tab is very useful. For understanding the program encoded in the weights of a neural net, it is not very useful, I want other tools.
7:50 PM - 11 May 2020
0 replies
0 retweets
1 like
Loading seems to be taking a while.
Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.