Abhirath Anand

Differentiating through this is what causes the error for me: ```julia function (m::DenseBlock)(x) input = [x] for layer in m.layers x = layer(input) input = vcat(input, [x]) end return cat_channels(input...)...

I've had less time to work on this in the recent past, but I'm going to try and push some of these refactors through in the next few months. However,...

`inchannels` and `nclasses` have been added as kwargs to all the models, I believe. But `imsize` is not so straightforward, so support for that has not yet landed for all...

> I'm not sure if this warrants another issue but the docstrings for models without `imsize` also don't say what the expected imsize is. Oops, I'll try and work on...

I was trying a lot of these on CIFAR-10 during my GSoC and was facing issues, including with ResNet – the accuracies were nowhere near what I could get with...

> Hi, I know it is sort of a hot-fix to get MarkText running, but the following command did the trick for me on an M1 MBA running macOS v12.1:...

Can confirm the same behaviour on Flux v0.13.6 on the GPU

Hey, sorry I've been inactive for some time, college has been a little busy! I can definitely rebase this PR. The reason I haven't actually just finished this up is...

Given that activation functions have always been handled weirdly, despite the fact that it is a little against the Flux style, maybe it might not be a bad idea to...

This is what I get: ```julia julia> @time_imports using MLUtils 10.4 ms ┌ MacroTools 17.8 ms ┌ ZygoteRules 34.58% compilation time 0.6 ms ┌ DefineSingletons 0.3 ms ┌ IteratorInterfaceExtensions 0.9...