add instructions regarding jlab

Author: ericmjl

README.md

@@ -21,6 +21,17 @@ If you want `jax` with GPU, you will need to build from source, or follow the [i
 
 [![Binder](https://mybinder.org/badge_logo.svg)](https://mybinder.org/v2/gh/ericmjl/dl-workshop/master)
 
+### Notes
+
+If you are using Jupyter Lab, you will want to also ensure that `ipywidgets` is installed:
+
+```bash
+# only if you don't have ipywidgets installed.
+$ conda install -c conda-forge ipywidgets
+# the next line is necessary.
+$ jupyter labextension install @jupyter-widgets/jupyterlab-manager
+```
+
 ## Key Ideas
 
 The key idea for this tutorial is that if we really study deep learning's fundamental model, linear regression, then we can get a better understanding of the components - a model with parameters, a loss function, and an optimizer to change the parameters to minimize the loss. Most of us who become practitioners (rather than researchers) can then take for granted that the same ideas apply to any more complex/deeper model.