neuronika VS dfdx

Also perhaps comparing to Neuronika.

how does it compare with https://github.com/spearow/juice, https://github.com/neuronika/neuronika and https://github.com/spearow/juice?

Check where and how this struct is used. https://github.com/neuronika/neuronika/blob/variable-rework/neuronika-variable/src/history.rs

Also you can take a look at neuronika.

There is also https://github.com/neuronika/neuronika

I have a question: as the maintainer of [neuronika](https://github.com/neuronika/neuronika), a crate that offers dynamic neural network and auto-differentiation with dynamic graphs, I'm looking at a future possible feature for such framework consisting in the possibility of compiling models, getting thus rid of the "dynamic" part, which is not always needed. This would speed the inference and training times quite a bit.

If you like autograd you will love neuronika

After reading most of the book, I wanted to get my hands dirty. My initial idea was to build a small machine learning framework but I deemed it to be too difficult if not impossible for me at the time. (Now, neuronika would have something to say). When gathering the bibliography for my thesis, I recalled to have stumbled upon a particular algorithm, t-SNE, whom I liked very much. I found the idea behind it to be very clever and elegant (t-SNE it's still one of my favorite algorithms, together with backprop and SOM, I find manifold learning fascinating in general). "So be it", I said, and I began writing a mess of a code, that was basically a translation of the C++ implementation. Boy was it bad.

I keep checking the progress on dfdx for this reason. It does what I (and, I assume from context, you) want: Provides static checking of tensor shapes. Which is fantastic. Not quite as much inference as I'd like but I love getting compile-time errors that I forgot to transpose before a matmul.

It depends on the generic_const_exprs feature which is still, to quote, "highly experimental":

https://github.com/rust-lang/rust/issues/76560

Definitely not for production use, but it gives a flavor for where things can head in the medium term, and it's .. it's nice. You could imagine future type support allowing even more inference for some intermediate shapes, of course, but even what it has now is really nice. Like this cute little convnet example:

https://github.com/coreylowman/dfdx/blob/main/examples/night...

And for transformers, it's really early days for dfdx, but it's a library that aims to sit basically at the Pytorch level of abstraction, that the difference is it's not just coded in Rust, but it follows the Rust-y/functional-y philosophy of "if it compiles it runs".

Wow that is super interesting. I actually tried to use GATs at first to be generic over shapes, but I couldn't do it, I'm sure it would be possible in the future though. There is this library dfdx that does something similar to what you mentioned, but it feels a little clumsy to me.

Awesome, I added an issue here https://github.com/coreylowman/dfdx/issues/597. We can discuss more there! The first step will just be adding the device and implementing tensor creation methods for it.

The next release of dfdx includes a CUDA device and implements many ops. The same dev created a new crate, cudarc, for a wrapper around CUDA toolkit.

A question I have is: what are the philosophical/design differences with dfdx? As someone who's played around with dfdx and only skimmed the README of burn, it seems like dfdx leans into Rust's type system/type inference for compile time checking of as much as is possible to check at compile time. I wonder if you've gotten a chance to look at dfdx and would like to outline what you think the differences are. Thanks!