habedi0

Lol

Looks like it's over for Linux.

`rustworkx` is older and much more mature than PyGraphina. So at the moment, it includes a larger collection of graph algorithms. But the goal is to keep PyGraphina focused on specific applications like community detection and link prediction with a high-level API like NetworkX.

I wanted to get a sense of and intuition for how to build an LLM from scratch, and to learn more Zig by doing so.

I avoided using external (linear algebra/tensor) libraries to keep the project's scope small and manageable. Adding them can be the next step, but they are usually very large dependencies that can make the project bloated. Anyway, Zig has great SIMD and multi-threading support, but I think there is a need for a native linalg/tensor library for Zig with a clean and mature API.

I'm not sure about the usefulness, it depends TBH. I think there are a lot of people (like hobbyists, students, people learning Zig, etc.) who can find the project somewhat useful at its current stage.

I picked that name (`ordered`, as in `put into and kept in an arrangement`) when I started the project a few months ago and decided to keep it.

This is pretty cool!

That's kinda weird and sad. Lambdas are very useful TBH.

I think they want to add lambdas to Zig at some point. The number of changes is high between releases.

I created this library to help me verify the correctness of my Zig code in another project. However, I wanted the option to strip the checks when the code becomes more mature, for efficiency. For efficiency, I mainly care about speed, which is why checks are only compiled out in `ReleaseFast`.

Thanks for the share.

Thanks.

I'm new to Zig myself and created this library because I had trouble verifying the correctness of the code I wrote for a set of complex data structures in another larger Zig project (this one: https://github.com/habedi/ordered). I'm currently experimenting with how to use DbC in Zig, which led to the creation of this library.

Looks like a great (offline) use case for small language models. Are you going to share the source code for your implementation?

The Story of Mel is all you need?

Hi, sure.

This is pretty cool!

Thanks.

Lol. No. Not heretical at all.

I might add support for other model providers (like Google and Azure) in the future. Although I'm trying to keep the scope of the project very small because it's easier for me to maintain it.

Anyway, I think adding new LLM providers is pretty straightforward if you want to do it yourself. You just need to implement the API of the BaseLLM (see the `cogitator/model/base.py` file) for your provider. After that, you just use it like how you use OllamaLLM or OpenAILLM.

I'm not using it in a larger project at the moment. The examples right now are mainly included to help people get started quickly. About the applications, they are somewhat context-dependent, but I might add one or two larger examples later if I have the time.

I guess CoT prompting could be used for ARC Prize (https://arcprize.org)

Sounds like a great idea. In the next release, I'll add a visualization for how things work and related to each other, and possibly also include some benchmark results.

Lol. No

Hi everyone,

I'm developing an open-source Rust library called Graphina for graph data science. The library is in a very early stage of development, so it may not be ready for serious use yet. However, early versions of many core features and algorithms have already been implemented.

I'm announcing the project here to invite contributions and suggestions from the community. I'm still relatively new to Rust, so I would especially appreciate any constructive feedback you might have to help improve Graphina.

GitHub link of the project: https://github.com/habedi/graphina

Thanks for your interest and support!