Hacker News: megadragon9

Self-Improving Harness Is an Experiment Design Problem

megadragon9 — Sat, 13 Jun 2026 01:50:26 +0000

Article URL: https://www.henrypan.com/blog/2026-05-25-self-improvement-harness/

Comments URL: https://news.ycombinator.com/item?id=48511641

Points: 2

# Comments: 0

New comment by megadragon9 in "Google to pay SpaceX $920M a month for compute capacity at xAI data centers"

megadragon9 — Sat, 06 Jun 2026 20:11:33 +0000

looks like elon web services (EWS) is the master plan all along :D

New comment by megadragon9 in "Can the stockmarket swallow Anthropic, SpaceX and OpenAI?"

megadragon9 — Tue, 02 Jun 2026 01:52:21 +0000

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" essay covered it well [0]

[0]: https://www.owenmcgrann.com/p/the-dead-economy-theory

Show HN: What 1k Harness Experiments Taught Me About Self-Improving Agents

megadragon9 — Thu, 28 May 2026 01:47:27 +0000

Article URL: https://www.henrypan.com/blog/2026-05-25-self-improvement-harness/

Comments URL: https://news.ycombinator.com/item?id=48303342

Points: 3

# Comments: 0

New comment by megadragon9 in "What 1k Harness Experiments Taught Me About Self-Improving Agents"

megadragon9 — Wed, 27 May 2026 17:05:19 +0000

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 weeks, I think the continuous self-improvement is mostly an experiment-systems problem. The system needs a way to decide what kind of improvements can safely compound.

Turns out there's a lot of parallels to coding-agent customization (e.g. SKILLS.md etc..) too.

I wrote my experience of building such system here, including the successful and failure attempts during the process, and how I approached the self-improvement loop. It's not intended as a benchmark claim but more of a systems/research writeup.

https://www.henrypan.com/blog/2026-05-25-self-improvement-ha...

What 1k Harness Experiments Taught Me About Self-Improving Agents

megadragon9 — Wed, 27 May 2026 17:05:19 +0000

Article URL: https://www.henrypan.com/blog/2026-05-25-self-improvement-harness/

Comments URL: https://news.ycombinator.com/item?id=48297185

Points: 2

# Comments: 1

New comment by megadragon9 in "Ask HN: What are you working on? (May 2026)"

megadragon9 — Fri, 15 May 2026 06:50:21 +0000

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 good learning experience to work without all the high-level abstractions to "build a wheel" and "use that wheel to build a car". Post-training results are still cooking, since training on my MacBookPro is quite slow with "unoptimized PyTorch" :)

1. https://github.com/workofart/ml-by-hand

2. https://github.com/workofart/ml-by-hand/blob/main/examples/s...

3. https://github.com/workofart/ml-by-hand/blob/main/examples/g...

How a Deep Learning Library Enables Learning

megadragon9 — Mon, 16 Mar 2026 23:40:59 +0000

Article URL: https://www.henrypan.com/blog/2026-03-14-how-deep-learning-library-enables-learning/

Comments URL: https://news.ycombinator.com/item?id=47406602

Points: 2

# Comments: 0

How A Deep Learning Library Enables Learning

megadragon9 — Sun, 15 Mar 2026 17:52:50 +0000

Article URL: https://www.henrypan.com/blog/2026-03-14-how-deep-learning-library-enables-learning/

Comments URL: https://news.ycombinator.com/item?id=47389887

Points: 2

# Comments: 0

New comment by megadragon9 in "Claude March 2026 usage promotion"

megadragon9 — Sat, 14 Mar 2026 22:42:24 +0000

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 spectrum towards more "real-time" demand shaping.

New comment by megadragon9 in "Build a Deep Learning Library"

megadragon9 — Fri, 02 Jan 2026 01:05:32 +0000

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] from scratch, using nothing but Python and NumPy. I started with a tiny autograd engine, then gradually created layer modules, optimizers, data loaders etc... I simply wanted to learn machine learning from first principles. Along the way I attempted to reproduce classical convnets [2] all the way to a toy GPT-2 [3] using the library I built. It definitely helped me understand how machine learning worked underneath the hood without all the fancy abstractions that PyTorch/TensorFlow provides. I eventually wrote a blog post [4] of this journey.

[1] https://github.com/workofart/ml-by-hand

[2] https://github.com/workofart/ml-by-hand/blob/main/examples/c...

[3] https://github.com/workofart/ml-by-hand/blob/main/examples/g...

[4] https://www.henrypan.com/blog/2025-02-06-ml-by-hand/

New comment by megadragon9 in "NanoChat – The best ChatGPT that $100 can buy"

megadragon9 — Fri, 17 Oct 2025 05:12:21 +0000

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]. Learning is done by "re-inventing the wheel" yourself, one step at a time :)

[1] https://github.com/workofart/ml-by-hand

New comment by megadragon9 in "Claude Code Pro Limit? Hack It While You Sleep"

megadragon9 — Sun, 06 Jul 2025 19:19:36 +0000

Reminds me of this HN discussion (Writing Code Was Never the Bottleneck): https://news.ycombinator.com/item?id=44429789

New comment by megadragon9 in "Reinvent the Wheel"

megadragon9 — Sat, 24 May 2025 20:43:23 +0000

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 scratch, using nothing but Python and NumPy. I started with a tiny autograd engine, then gradually created layer modules, optimizers, data loaders etc... I simply wanted to learn machine learning from first principles. Along the way I attempted to reproduce classical convnets [2] all the way to a toy GPT-2 [3] using the library I built. It definitely helped me understand how machine learning worked underneath the hood without all the fancy abstractions that PyTorch/TensorFlow provides. Kinda like reinventing the car using the wheel I reinvented :)

[1] https://github.com/workofart/ml-by-hand

[2] https://github.com/workofart/ml-by-hand/blob/main/examples/c...

[3] https://github.com/workofart/ml-by-hand/blob/main/examples/g...

New comment by megadragon9 in "The Llama 4 herd"

megadragon9 — Sat, 05 Apr 2025 18:52:50 +0000

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

New comment by megadragon9 in "Ask HN: What are you working on? (March 2025)"

megadragon9 — Sun, 30 Mar 2025 21:33:40 +0000

I built a machine learning library [1] (similar to PyTorch's API) entirely from scratch using only Python and NumPy. It was inspired by Andrej Karpathy's Micrograd project [2]. I slowly added more functionality and evolved it into a fully functional ML library that can build and train classical CNNs [3] to even a toy GPT-2 [4].

I wanted to understand how models learn, like literally bridging the gap between mathematical formulas and high-level API calls. I feel like, as a beginner in machine learning, it's important to strip away the abstractions and understand how these libraries work from the ground up before leveraging these "high-level" libraries such as PyTorch and Tensorflow. Oh I also wrote a blog post [5] on the journey.

[1] https://github.com/workofart/ml-by-hand

[2] https://github.com/karpathy/micrograd

[3] https://github.com/workofart/ml-by-hand/blob/main/examples/c...

[4] https://github.com/workofart/ml-by-hand/blob/main/examples/g...

[5] https://www.henrypan.com/blog/2025-02-06-ml-by-hand/

New comment by megadragon9 in "Write to Escape Your Default Setting"

megadragon9 — Fri, 28 Feb 2025 22:06:03 +0000

> Writing reveals what you don’t know, what you can’t see when an idea is only held in your head. Biases, blind spots, and assumptions you can’t grasp internally.

I completely agree with this. Often, I think I understand something, but when I try to explain it to others, I quickly realize where my understanding is shaky. The gaps become even more apparent when I attempt to write it down because I have to structure my thoughts logically and precisely. Writing goes a step beyond speaking because it forces me to re-read and refine my ideas, whereas spoken words often disappear without deeper reflection. Oh, even this comment that I'm writing now was edited a couple of times before submitting it. The second half of the comment was added after re-reading the first half.

Understanding ML: Creating a PyTorch-Inspired Deep Learning Library from Scratch

megadragon9 — Sat, 22 Feb 2025 18:42:41 +0000

Article URL: https://www.henrypan.com/blog/2025-02-06-ml-by-hand/

Comments URL: https://news.ycombinator.com/item?id=43141751

Points: 1

# Comments: 0

New comment by megadragon9 in "Show HN: Transform your codebase into a single Markdown doc for feeding into AI"

megadragon9 — Fri, 14 Feb 2025 18:25:54 +0000

I would say the demand for this kind of tool definitely exists. Good work! From a rough glance it looks pretty similar to another tool that I've been using https://github.com/mufeedvh/code2prompt

Show HN: ML Library created by Python+NumPy (trains CNNs to a toy GPT-2)

megadragon9 — Wed, 12 Feb 2025 02:07:31 +0000

Hi HN, I built a machine learning library entirely from scratch using only Python and NumPy. I then used it to train a range of models—from classical CNNs, ResNets, RNNs, and LSTMs to modern Transformers and even a toy GPT-2. The motivation came from my curiosity about how to build deep learning models from scratch, like literally from mathematical formulas. I built this project not to replace production-ready libraries like PyTorch or TensorFlow, but to strip away the abstractions and reveal the underlying mathematics of machine learning.

Key points:

- Everything is derived in code — no opaque black boxes.

- API mirrors PyTorch so you can pick it up quickly.

- You can train CNNs, RNNs, Transformers, and even GPT models.

- Designed more for learning/debugging than raw performance.

What’s different here?

While there are many powerful ML libraries available (TensorFlow, PyTorch, Scikit-learn, etc.), they often hide the underlying math behind layers of abstraction. I believe that to truly master these tools, you first need to understand how they work from the ground up. This project explicitly derives all the mathematical and calculus operations in the code, making it a hands-on resource for deepening the understanding of neural networks and library building :)

Check it out:

- Github Repository: https://github.com/workofart/ml-by-hand

- API Documentation: https://ml-by-hand.readthedocs.io/en/latest/

- Examples: Explore models like GPT-2, CNNs, Transformers, and LSTMs in the examples/ folder: https://github.com/workofart/ml-by-hand/tree/main/examples

- Blog Post: Read about the project’s motivation, design, and challenges at https://www.henrypan.com/blog/2025-02-06-ml-by-hand/

I’d love to hear any thoughts, questions, or suggestions — thanks for checking it out!

Comments URL: https://news.ycombinator.com/item?id=43020982

Points: 1

# Comments: 0