Data Science

  • 3 Posts
  • 20 Comments
Joined 2 years ago
cake
Cake day: June 17th, 2023

help-circle
  • Introduction

    I’ve written quite a lot of free software in my life. Most of it was from scratch: projects I started myself. So I get to choose where to host them – or rather, I have to choose where to host them.

    These days, all my projects are held in Git. And mostly, I put them in ‘bare’ git repositories on my personal website.

    I don’t use any git ‘forge’ system layered on top of Git, like Gitlab or Github, which automatically makes a bug tracking database for each project, and provides a convenient button for a user to open a merge request / pull request. I just use plain Git. People can ‘git clone’ my code, and there’s a web-based browsing interface (the basic gitweb) for looking around without having to clone it at all. But that’s all the automated facilities you get.

    Occasionally this confuses people, so I thought I should write something about it.

    Discussion with the author @ https://hachyderm.io/@simontatham/114111520633445984













  • From the article:

    DeepSeek-R1 release leaves open several questions about:

    • Data collection: How were the reasoning-specific datasets curated?
    • Model training: No training code was released by DeepSeek, so it is unknown which hyperparameters work best and how they differ across different model families and scales.
    • Scaling laws: What are the compute and data trade-offs in training reasoning models?

    These questions prompted us to launch the Open-R1 project, an initiative to systematically reconstruct DeepSeek-R1’s data and training pipeline, validate its claims, and push the boundaries of open reasoning models. By building Open-R1, we aim to provide transparency on how reinforcement learning can enhance reasoning, share reproducible insights with the open-source community, and create a foundation for future models to leverage these techniques.

    In this blog post we take a look at key ingredients behind DeepSeek-R1, which parts we plan to replicate, and how to contribute to the Open-R1 project







  • Embedded software development has dramatically advanced over the past decade. What does that mean for bare-metal programming?

    At a Glance

    • Bare-metal programming is an essential skill as it enables you to understand what your system is doing at the lowest levels.
    • Even if you spend your days working with abstraction layers, bare-metal programming will guide you should abstractions fail.
    • And bare-metal skills can provide a solid foundation for troubleshooting and debugging.