Juheon Lee

No problem at all :) Thanks for your kind explanation. My concern was that the current implementation of the ``stax.Aggregate`` only accepts dense graphs, thus memory costs increase quadratically with...

Does neural-tangent support element-wise product? I encountered a similar problem. In my case, I want to do element-wise product instead of subtraction. something like this ```python input1 = Input(shape=(n,3), dtype='float32',...

Thanks for your help! That's exactly what I want to do. Perhaps it may not make much sense for the standard neural networks. But NTK (thanks to its covariance form)...

applying pyramid is very straightforward. You need to change scale the input image with imresize. I am happy to implement it, will update soon.

 Excluding entry point one, there are 3 more errors. 1. Checkpointing is not compatible with .grad() or when an `inputs` parameter is passed to .backward(). This one is because...