Abhirath Anand

Just for tracking, `gradtests` on Metalhead (Julia 1.8): 

> I think I missed some of the discussion on this, what do the top and bottom test runs represent? Is the bottom just a second run after warmup or...

Is there a way to toggle that particular option in the configuration file?

@darsnack @ToucheSir what do you make of this proposal? This might reduce code duplication if we could write the methods for just the `MetalheadModel` type (not to mention that this...

> though I think you really only remove the forward pass and maybe `@functor`? Also `backbone` and `classifier`. > What I want to avoid is any notion that a user...

Funny, I can't seem to remove `@functor` when I try this. It errors by saying `type does not have a definite number of fields`?

I'd love to help in whatever way I can....is there a reason that Metalhead doesn't have DataAugmentation built into it though? I've noticed that the Julian way is to keep...

> We could make a FluxVision.jl umbrella Even if not in terms of packages, maybe I could come up with docs that document the most common usecases - but TBVH,...

I'm planning on - [ ] [CaiT](https://arxiv.org/abs/2103.17239) - [ ] [LeViT](https://arxiv.org/abs/2104.01136) - [ ] [Swin](https://arxiv.org/abs/2103.14030) - [ ] [MobileViT](https://arxiv.org/abs/2110.02178) - [x] [ConvNeXt](https://arxiv.org/abs/2201.03545) at first. I also want to start on...

`nclasses` is supported by every current Metalhead model, I believe. `inchannels` isn't but I have been working on trying to unify the API a little bit (see #174 for ResNet,...