[D] What are people here using to visualize gradient flow / distribution of activations in their models?

What are the good tools for checking things like the following:

\- How fast the gradients for each layer in your model are updating (useful to see which layers are learning fast, which layers might not be learning fast)

\- Visualizing the stdev of activations / the magnitude of the weights

Other such things, which help give a high-level overview of how one's model is doing.

distribution flow gradient machinelearning people