equinox VS serenity

I wrote a JAX-based neural network library (Equinox [1]) and numerical differential equation solving library (Diffrax [2]).

At the time I was just exploring some new research ideas in numerics -- and frankly, procrastinating from writing up my PhD thesis!

But then one of the teams at Google starting using them, so they offered me a job to keep developing them for their needs. Plus I'd get to work in biotech, which was a big interest of mine. This was a clear dream job offer, so I accepted.

Since then both have grown steadily in popularity (~2.6k GitHub stars) and now see pretty widespread use! I've since started writing several other JAX libraries and we now have a bit of an ecosystem going.

[1] https://github.com/patrick-kidger/equinox

The elevator pitch is Optimistix is really fast, especially to compile. It plays nicely with Optax for first-order gradient-based methods, and takes a lot of design inspiration from Equinox, representing the state of all the solvers as standard JAX PyTrees.

If you like PyTorch then you might like Equinox, by the way. (https://github.com/patrick-kidger/equinox ; 1.4k GitHub stars now!)

You're thinking of `jax.closure_convert`. :)

(Although technically that works by tracing and extracting all constants from the jaxpr, rather than introspecting the function's closure cells -- it sounds like your trick is the latter.)

When you discuss dynamic allocation, I'm guessing you're mainly referring to not being able to backprop through `jax.lax.while_loop`. If so, you might find `equinox.internal.while_loop` interesting, which is an unbounded while loop that you can backprop through! The secret sauce is to use a treeverse-style checkpointing scheme.

https://github.com/patrick-kidger/equinox/blob/f95a8ba13fb35...

I'm a big fan of using ABCs to declare interfaces -- so much so that I have an improved abc.ABCMeta that also handles abstract instance variables and abstract class variables: https://github.com/patrick-kidger/equinox/blob/main/equinox/_better_abstract.py

For the daily research, I use Equinox (https://github.com/patrick-kidger/equinox) as a DL librarry in JAX.

It sounds like you're concerned about how downstream libraries tend to wrap JAX transformations to handle their own thing? (E.g. `haiku.grad`.)

If so, then allow me to make my usual advert here for Equinox:

https://github.com/patrick-kidger/equinox

This actually works with JAX's native transformations. (There's no `equinox.vmap` for example.)

On higher-order functions more generally, Equinox offers a way to control these quite carefully, by making ubiquitous use of callables that are also pytrees. E.g. a neural network is both a callable in that it has a forward pass, and a pytree in that it records its parameters in its tree structure.

SerenityOS replicates that look and feel. It is also implemented in a dialect of C++ that adheres to some of the good parts of C++98: https://serenityos.org

One example of a useful technique

https://serenityos.org/ apparently only makes source code available. There are no binary images of the OS to install

I think Andreas said this functions like a little test -- if you're not willing to build it from source, then you probably wouldn't be a good contributor anyway.

---

Likewise, my shell project provides source tarballs only, right now - https://www.oilshell.org/release/0.21.0/

It is packaged in a number of places, which I appreciate. That means some other people are willing to do some work.

And they provide good feedback.

I would like it to be more widely available, but yeah I definitely see that you need to "gate" peanut gallery feedback a bit, because it takes up a lot of time.

Of course, it's a tricky balance, because you also want feedback from casual users, to make the project better.

Indeed, given the existence of `JS::NonnullGCPtr`, `JS::GcPtr` intentionally corresponds to a nullable pointer, so it seems dangerous to convert one to a reference without a null-check.

That said, a naive code search finds what *may* be more cases of this pattern:

https://github.com/search?q=repo%3ASerenityOS%2Fserenity+%2F...

Eg: https://github.com/SerenityOS/serenity/blob/a68b134e6dea5065... -> https://github.com/SerenityOS/serenity/blob/a68b134e6dea5065...

In some of those search results, it is fine because there is a preceding null-check, and obviously I know nothing about this code other than this naive search result, but perhaps it would be prudent to vet all of them.

It is a SerenityOS project. You can find the answer to that question in their primary project's FAQ[1].

1. https://github.com/SerenityOS/serenity/blob/master/Documenta...

https://github.com/SerenityOS/serenity

The best way to write proper exception free C++ is not to use the C++ Standard Library.

I initially confused it with Serenity OS prior to watching the video: https://github.com/SerenityOS/serenity

My contributions to SerenityOS[0] helped me get my current job. My team lead (who was also my interviewer) was interested in what I did since I listed some of it in my CV, and I showed him some PRs I made and explained what went into each of them. It was really exciting because I didn't have professional experience with low-level development, and basically got the job due to hobby programming.

[0]: https://github.com/SerenityOS/serenity/pulls?q=is%3Apr+autho...

Definitely not "literally impossible", just a great deal of work. https://github.com/SerenityOS/serenity/tree/master/Ladybird