muragekibicho

8 ms·

1. 61.▲
   by muragekibicho 10mo ago

   Trump defunded my math PhD so I got a programmer job just to pay the bills. The Substack lets me do what I love. Thanks for this comment!
2. 62.▲
   by muragekibicho 10mo ago

   I run a Substack where I show programmers how to turn advanced math papers into C and Python code. I'm a math guy who codes and I do it for fun. I'm shocked people are interested in this stuff. LeetArxiv Substack: https:/&#x
3. 63.▲
   by muragekibicho 10mo ago

   I have a laptop with 4 GB of ram and firefox keeps crashing. I wish they'd fix this instead of saddling me with AI features I don't need.
4. 64.▲
   by muragekibicho 10mo ago

   Is it an inside joke I missed? 'Militarily' here and another comment had 'Bigly'. Is it a Trumpism?
5. 65.▲
   Tell HN: GitHub listened and removed the AI summary feature

   3 points by muragekibicho 10mo ago|0 comments
6. 66.▲
   by muragekibicho 10mo ago

   I wish they focused on making a GoogleColab alternative where we could use Tenstorrent hardware. They're pushing towards devs but cite not a dearth of accessible devices. I have neither an Nvidia GPU nor a Google TPU but guess what: I&
7. 67.▲
   by muragekibicho 10mo ago

   You made me look 'blog//2025-12-05/leaving-intel.html'
8. 68.▲
   by muragekibicho 11mo ago

   What are the odds. OP's rant is about Helmet.js, his sorta-popular NPM package. You drew a fascinating parallel.
9. 69.▲
   by muragekibicho 11mo ago

   I share your sentiment. He needs to decide if his software is open-source or not. It goes both ways. Open-source devs don't owe you free bug patches. People profitting from open-source don't owe you a share of their income.
10. 70.▲
    by muragekibicho 11mo ago

    VS Code is the chromium of IDEs. I shall not explain further.
11. 71.▲
    by muragekibicho 11mo ago

    'or if reading is too woke' Amazing piece and oddly relatable
12. 72.▲
    by muragekibicho 11mo ago

    Gaudry-Schost is a lesser-known alternative to Pollard Rho for solving discrete logarithms. The authors found an interesting alternative to the Birthday Paradox: If we have 365 balls and draw them with replacement, then record the picked ba
13. 73.▲
    by muragekibicho 11mo ago

    What's the criteria for changing HN post titles? The earlier posting was 'A marriage proposal hidden across two playstation games' or something like that. Is there a HN bot doing this or is it the mods?
14. 74.▲
    by muragekibicho 11mo ago

    Exactly where my mind turned. It's interesting how the AI OG's (Feifei and Cunn) think world models are the way forward.
15. 75.▲
    by muragekibicho 11mo ago

    Honestly, I thought the diatribe would focus on needless complexity. The starting example is how I'd do it in C: ``` void f(const char* p) // unsafe, naive use { FILE \*f = fopen(p, "r"); /&#
16. 76.▲
    by muragekibicho 11mo ago

    They haven't and that's the crux of it all: they can sue if they want or when they want.
17. 77.▲
    by muragekibicho 11mo ago

    Honestly, I don't even know where to begin. It's insane IBM owns the patent to continued fractions. If you wrote a continued fraction class in Pytorch and called backwards (or even differentiated the power series) then you're
18. 78.▲
    by muragekibicho 11mo ago

    Here it is: https://patents.justia.com/patent/20230401438 On Google Patents: https://patents.google.com/patent/US20230401438A1/en The authors simply implement a continued fraction library in
19. 79.▲
    by muragekibicho 11mo ago

    I'm working on fighting IBM's patent trolls. IBM slapped the words 'AI Interpretability' on Gauss' 200 year old continued fractions and was awarded a patent. Now they can charge rent if they encounter a continued f
20. 80.▲
    by muragekibicho 11mo ago

    I've learnt this the hard way. The most important thing is to get you to click. Sometimes I'll first iterate over the title before even writing on substack.
21. 81.▲
    by muragekibicho 11mo ago

    I googled Taboola and Outbrain. I bet you used 'Taboola or Outbrain' in the pejorative but I can't tell exactly what's wrong with the two companies lol.
22. 82.▲
    by muragekibicho 11mo ago

    Shameless plug. I run LeetArxiv. It's a successor to papers with code built on the thesis "Every research paper should be a brief blog post with relevant code". Here's our "Sora's Annotated Diffusion Transforme
23. 83.▲
    by muragekibicho 11mo ago

    Lol is this the Hacker News version of "I am Groot". Sidenote, I attempted a startup in the "Google Colab for GPUs" space. You approach won't work unless you give a generous free tier and bankroll content creators t
24. 84.▲
    by muragekibicho 1y ago

    Paper Summary: They replaced the U-net in a diffusion model with a Transformer. We coded this paper in Pytorch, it runs on CPU
25. 85.▲
    by muragekibicho 1y ago

    Lots of model releases are like this. We can only upvote. We can't run the model on our personal computers. We can neither test their 'Efficient Attention' concept on our personal computers. Honestly, it would take 24 hours
26. 86.▲
    by muragekibicho 1y ago

    I've not yet published the writeup. Only the C code is available atm. I solo run a thing called LeetArxiv. It's a successor to Papers with Code since the latter shut down.
27. 87.▲
    by muragekibicho 1y ago

    I coded the paper Sinkhorn Solves Sudoku. Lol the most bizarre algorithms can solve sudoku. https://github.com/MurageKibicho/Sinkhorn-Solves-Sudoku
28. 88.▲
    by muragekibicho 1y ago

    SATs are cool but MILPs are cooler IMO. Lol I've been trying to train a neural network over a finite field, not the reals and oh my god MILPs are God's gift to us.
29. 89.▲
    by muragekibicho 1y ago

    Isn't your diatribe contradictory. Your last paragraph appears to contradict your 'beliefs'.
30. 90.▲
    by muragekibicho 1y ago

    The text version of the 'airplane with bullet holes' meme lol