🚀 Community Collaboration: Let's Set Up a SETI-like Resource Sharing System for MLX Projects! 🤝

[JosefAlbers]

Hey MLX community!

I wanted to propose an idea that could greatly benefit those of us working on resource-intensive open source projects. We all know that sometimes our machines just don’t have the power or time to handle extensive computations—like training models or running long processes.

I’m suggesting we create a SETI-like resource sharing system where those with spare Mac machines can help run code that requires more resources. For example, I’m interested in implementing and experimenting with new models and methods published in AI/ML. These tasks can often take days to complete, especially when working with larger datasets.

One key advantage of this approach is that we can run different variations of a model or tweak parameters in parallel. This would allow us to get results more quickly and accelerate our iterative development cycles.

Many of us experience periods where our Macs are idle or only handling light tasks, such as streaming music playlists on YouTube. During those short bursts of heavy usage however, we often hit our RAM limits. This system could help us make better use of those idle moments while supporting each other’s projects.

Here’s how we could set this up:

1. Form a List: Create a shared document or channel where members can sign up to offer their machines and specify the resources they can provide (RAM, processing power, etc.).

2. Request System: If you need help, post your project details, including what you need and how long it might take. Those with spare resources can volunteer to run it.

3. Feedback Loop: After tasks are completed, we can create a way for both the requester and the helper to share feedback, ensuring a smooth and respectful exchange.

4. Community Guidelines: Establish some basic guidelines to keep everything organized and fair.

As an example, here are my Macs’ specs:

• MacBook Air (2022)

  • Processor: Apple M2
  • RAM: 8 GB
  • Storage: 256 GB SSD

• Mac Studio (2022)

  • Processor: Apple M1 Max
  • RAM: 64 GB
  • Storage: 512 GB SSD

This could be a fantastic way to foster collaboration, speed up our projects, and share resources effectively. Let me know your thoughts and if you’re interested in participating!

Looking forward to hearing from you!