Skip to content

Latest commit

 

History

History
27 lines (16 loc) · 2.3 KB

continue-working-for-the-hurd.md

File metadata and controls

27 lines (16 loc) · 2.3 KB
Raw

🚚 Continue Working for the Hurd

What is my vision?

I want to continue working for the GNU Hurd as a maintainer. However, I do not have too much confidence in managing memory manually by using C/C++.

This is a long-term target for me. Also, I am less experienced in Rust, I believe still some adapters I can go through with Rust for the Hurd.

And I see that Microsoft is rewriting parts of(36,000 lines) of the Windows kernel using the memory-safe programming language Rust. This is good news for me. As you can see that Rust focuses on safety and performance. It is a statically typed language, which means that the compiler can check for errors at compile time. This helps to prevent runtime errors, which are a major source of security vulnerabilities. And it is also a memory-safe language and brings Memory safety, Performance, and reliability these benefits for Windows Kernel.

Waiting for a good chance

Nowadays, more open-source AI projects are growing so fast. [Google's leaked file shows everything](https://github.com/orgs/community/discussions/54676). And more self-media on a social platform like Twitter "Sharing the latest developments in the world of artificial intelligence". And the breaking news is normal reports now.

So, I believe it is a good chance for me to do something on the Hurd. Because these techniques are totally easier than 5 years ago. At that time, I only had the `tensorflow` framework and no further resources. It is hard for me to do too many things. Now, I had more powerful tools that help me to do what I want to do.

About The Proposal

  1. I am working on the AI technique and trying to make the LLVMs more specific for my vision's tasks.
    1. The LoRA technique is a good starting point.
    2. Maybe I need to figure out which model is more effective for me.
    3. Re-training it by using promising techniques but are not limited to LoRA.
      1. And I want to do the re-training model task on the Steam Deck. We are all expected by using fewer resources to achieve more tasks.
  2. It is time to write a specific version of the copilot for myself or our team and it will be released as VSCode's plugin. It should be easy for me and others to access.
  3. After finishing No.2, it is time to make some effort on the GNU Hurd.