cargo-xtask VS neuronika

💡 Inspired by matklad/cargo-xtask and based on 🏃‍♂️ Write your Rust project scripts in task.rs from the Rust ecosystem.

Using a cargo-xtask task to generate them as a manual step (inlyne currently does this)

As someone else said just is good for that job, or you could implement an xtask helper for these things and setup a suitable development environment with that: https://github.com/matklad/cargo-xtask/

Well tools like cornucopia, prisma-rust-client, protoc-gen-tonic, they don't generate in build.rs, but instead provide either a cli to be called ahead of time, or provide a library that can be called by your own binary (which should generally follow the xtask pattern)

If I understand, this is a tool for when working on the project itself? Akin to a helper script? You could go the cargo install route as already pointed out but there is also the xtask convention.

For simple cli apps for internal use, such as cargo-xtasks, I prefer pico_args due to its fast compile times.

For developer-oriented stuff, there's tools like xshell and cargo-xtask. For operator tasks that need to run in a deployed environment, it's not usually a big lift to add CLI subcommands to your binary. It's certainly more boilerplate and inertia than doing stuff in a live REPL, though, and sometimes difficult to recommend for truly one-off situations.

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.