Broken validation; due to javascript error

[v-chanikya]

https://github.com/awslabs/amazon-sagemaker-examples/blob/5f187521b83a2aed1842b1a550afba90a08295a6/sagemaker_neo_compilation_jobs/mxnet_mnist/input.html#L48

when trying to run the validation by drawing on the canvas observed the following error in chrome debug console Uncaught ReferenceError: IPython is not defined at set_value (:48:22) at HTMLCanvasElement. (:22:29

Note: I am using sagemaker studio to run this example

This issue looks similar to the one describes @ https://github.com/jupyterlab/jupyterlab/issues/5888#issue-400974855

[fedeallocati]

This issue is still happening as of November 2021.

The Javascript API changed around the date of the issue linked (jupyterlab/jupyterlab#5888) and this example hasn't updated to use ipywidgets or another method for fetching the input drawing.

Note that this has an awful side effect on the notebook:

• The variable that Javascript is supposed to update is called data, which is the same variable name used for the training data.
• When the error that @v-chanikya mentioned happens and is logged in the debug console, the data variable is not overwriting, thus keeping the whole training data
• The next cell calls the endpoint for prediction using data. Since this is the whole dataset, after some time the kernel crashes trying to call the endpoint