neural-network-from-scratch VS Seed

I built a toy neural network that runs in the browser[1] to model 2D functions with the goal of doing something similar to this research (in a much more limited manner, ofc). Since the input space is so much more limited than language models or similar, it's possible to examine the outputs for each neuron for all possible inputs, and in a continuous manner.

In some cases, you can clearly see neurons that specialize to different areas of the function being modeled, like this one: https://i.ameo.link/b0p.png

This OpenAI research seems to be feeding lots of varied input text into the models they're examining and keeping track of the activations of different neurons along the way. Another method I remember seeing used in the past involves using an optimizer to generate inputs that maximally activate particular neurons in vision models[2].

I'm sure that's much more difficult or even impossible for transformers which operate on sequences of tokens/embeddings rather than single static input vectors, but maybe there's a way to generate input embeddings and then use some method to convert them back into tokens.

[1] https://nn.ameo.dev/

[2] https://www.tensorflow.org/tutorials/generative/deepdream

Full source code is on Github: https://github.com/ameobea/neural-network-from-scratch

Practically every Rust web frontend I've seen takes a react-like approach, with "hooks" to store all of the state in. The now-abandoned Seed and Yew's struct components use a message-passing approach, where the state is stored as member variables on the struct representing the component that are updated based on messages dispatched by event handlers. There's also egui, which has a completely different paradigm that involves making the UI from scratch every frame based on the app's current state. It's not a web framework the same way as the others, but it can draw its UI to a web canvas just fine.

Sycamore, Yew, or Seed if you want a full-stack solution. (Or Leptos if you want something that's faster but less mature.)

An authentication system is an integral part of modern applications. It's so important that almost all modern applications have some sort of it. Because of their critical nature, such systems should be secure and should follow OWAP®'s recommendations on web security and password hashing as well as storage to prevent attacks such as Preimage and Dictionary attacks (common to SHA algorithms). To demonstrate some of the recommendations, we'll be building a robust session-based authentication system in Rust and a complementary frontend application. For this article series, we'll be using Rust's actix-web and some awesome crates for the backend service. SvelteKit will be used for the frontend. It should be noted however that what we'll be building is largely framework agnostic. As a result, you can decide to opt for axum, rocket, warp or any other rust's web framework for the backend and react, vue or any other javascript framework for the frontend. You can even use rust's yew, seed or some templating engines such as MiniJinja or tera at the frontend. It's entirely up to you. Our focus will be more on the concepts.

If you want to do fullstack/SPA stuff, check out Sycamore, Seed, and Yew.

If you want to do front-end SPA development, take a look at Yew, Seed, or Sycamore.

Aside from Blazor there's already some other projects like Yew (rust), seed (rust), asm-dom (C++) and vugu (Go) and more that have decent followings and activity. A lot more (especially managed languages) are waiting for some features to come online like wasm GC and host bindings (direct wasm access to browser apis which includes the DOM). It'll take a bit of time, but it'll get there eventually.

To click * https://crates.io/crates/percy * https://crates.io/crates/seed * https://crates.io/crates/perseus * https://crates.io/crates/sycamore

For desktop apps maybe check out Tauri . You can use it with a lot of (web)frontend options including yew/wasm (also Seed ) if you want to go 100% Rust. Actix and Rocket are options for web framework. Also have look at the Building a Command Line Program in the book. I found it really helpful since i am just starting to learn myself.

I haven't been keeping up with it, but have heard of it. If ibrecall correctly it was created by the developer that initially developed seed (https://seed-rs.org/)