megadragon9

5 ms·

1. 1.▲
   by megadragon9 29d ago

   this makes me wonder, does having kids change the list of things on your list? and why?
2. 2.▲
   by megadragon9 1mo ago

   I think it's more about the scar tissue (a.k.a. intuition) when building LLMs from scratch than whether it's transferable to job search. Maybe the person will decide they don't like LLMs and not develop that into a career, or
3. 3.▲
   by megadragon9 1mo ago

   I did something similar but with smart home devices. I use the homebridge interface to connect my smart home devices to Apple's homekit protocol. Some homebridge plugins for my devices were outdated and no longer maintained, so I asked
4. 4.▲
   by megadragon9 2mo ago

   i created something similar a while back. Inspired by micrograd for showing the connection between math/calculus and code, then building along the way to a full NumPy deep learning library that I pretrained GPT-2 124M model with it. On
5. 5.▲
   Show HN: Auto-train the harness, not the LLM. cross-model, cross-benchmark gains(github.com)

   4 points by megadragon9 2mo ago|0 comments
6. 6.▲
   by megadragon9 2mo ago

   I did something similar. It started off as a "self-improving agent" project, inspired by autoresearch, then later on I reframed it as "harness training" (discrete program search) borrowing the mental model from ML traini
7. 7.▲
   Freeze the model, train the harness: gains transfer across LLMs and benchmarks(henrypan.com)

   2 points by megadragon9 2mo ago|0 comments
8. 8.▲
   Training Agent Harness Like Training a ML Model(henrypan.com)

   2 points by megadragon9 2mo ago|0 comments
9. 9.▲
   Show HN: Freeze the Model, Train the Harness(github.com)

   4 points by megadragon9 3mo ago|0 comments
10. 10.▲
    by megadragon9 3mo ago

    I worked on this project ( https://github.com/workofart/harness-training ) for the past few months to reframe "Agent-driven Self-improving Harness" to "Harness Training". The idea is simple, the harne
11. 11.▲
    Train a Harness to improve model/env-agnostic capabilities with PyTorch-like API(github.com)

    3 points by megadragon9 3mo ago|0 comments
12. 12.▲
    by megadragon9 4mo ago

    Interesting project. Do you think manual memory management help understand computational graph lifecycle better, or does it distract from backprop itself? btw, I went down the micrograd path with numpy-primitives all the way to building a P
13. 13.▲
    GPT-2 124M checkpoint pre-trained on OpenWebText 27.5B tokens(github.com)

    1 points by megadragon9 4mo ago|1 comments
14. 14.▲
    by megadragon9 4mo ago

    Model built and trained using a hand-built deep learning library (numpy primitives)
15. 15.▲
    by megadragon9 4mo ago

    I'm continuing to expand my own deep learning library [1] built with numpy-primitives to support LLM post-training techniques like supervised fine-tuning (SFT) and reinforcement learning with GRPO. It's a good learning experience
16. 16.▲
    Self-Improving Harness Is an Experiment Design Problem(henrypan.com)

    3 points by megadragon9 4mo ago|0 comments
17. 17.▲
    by megadragon9 4mo ago

    looks like elon web services (EWS) is the master plan all along :D
18. 18.▲
    by megadragon9 4mo ago

    I don't think the market will swallow the stock offerings until we see early signs of GDP growth attributable to these entities. But until then, I think the cost is higher than the benefit, which "The dead economy theory" ess
19. 19.▲
    Show HN: What 1k Harness Experiments Taught Me About Self-Improving Agents(henrypan.com)

    3 points by megadragon9 4mo ago|0 comments
20. 20.▲
    What 1k Harness Experiments Taught Me About Self-Improving Agents(henrypan.com)

    2 points by megadragon9 4mo ago|1 comments
21. 21.▲
    by megadragon9 4mo ago

    I recently wanted to see whether an AI agent could self-improve a harness to solve terminal bench tasks. It’s possible for an AI agent to propose a meaningful one-time change to the harness, but after experimenting with this for a couple of
22. 22.▲
    by megadragon9 5mo ago

    I'm continuing to expand my own deep learning library [1] (PyTorch-clone built with Python and Numpy) to support LLM post-training techniques like supervised fine-tuning (SFT) [2] and reinforcement learning with GRPO [3] . It's a
23. 23.▲
    How a Deep Learning Library Enables Learning(henrypan.com)

    2 points by megadragon9 7mo ago|0 comments
24. 24.▲
    How A Deep Learning Library Enables Learning(henrypan.com)

    2 points by megadragon9 7mo ago|0 comments
25. 25.▲
    by megadragon9 7mo ago

    Interesting to see more demand shaping mechanisms applied to LLM inference. Even though the "batch processing" feature is already available. I guess this "promotion" is to test the hypothesis of sliding along the spectru
26. 26.▲
    by megadragon9 9mo ago

    Thanks for sharing! It's inspiring to see more people "reinventing for insight" in the age of AI. This reminds me of my similar previous project a year ago when I built an entire PyTorch-style machine learning library [1] fro
27. 27.▲
    by megadragon9 1y ago

    Love the educational value of this "nano-sized" project. This reminded me of the from-scratch project I created to learn about deep learning libraries, neural networks all the way to LLMs like GPT-2 using just Numpy and Python [1]
28. 28.▲
    by megadragon9 1y ago

    Reminds me of this HN discussion (Writing Code Was Never the Bottleneck): https://news.ycombinator.com/item?id=44429789
29. 29.▲
    by megadragon9 1y ago

    Thanks for this inspiring essay, I couldn’t agree more that “reinventing for insight” is one of the best ways to learn. I had a similar experience couple months ago when I built an entire PyTorch-style machine learning library [1] from scra
30. 30.▲
    by megadragon9 1y ago

    The blog post is quite informative: https://ai.meta.com/blog/llama-4-multimodal-intelligence/