HNHacker News
TopNewBestAskShowJobs

dimatura

963 karma · joined May 22, 2010

submissionscomments
dimatura··on Path Isn't Real on Linux
And NFS!
dimatura··on A single line of code cost $8000
$8000 also seems pretty cheap for 2PB of traffic? Looking at google cloud storage egress rates, $0.02/GiB (which is on the lower end, since it depends on destination) would be about $40k for 2PB.
dimatura··on Watching 03 Model Sweat over a Paul Morphy Mate-in-2
Honestly, I think that if in 2020 you had asked me whether we would be able to do this in 2025, I would've guessed no, with a fairly high confidence. And I was aware of GPT back then.
dimatura··on TmuxAI: AI-Powered, Non-Intrusive Terminal Assistant
The "non-intrusive" part is interesting. I've bit the bullet with AI assistance when coding - even when it feels like it gets in the way sometimes, overall I find it a net benefit. But I briefly tried AI in the shell with the warp terminal and found it just too clunky and distracting. I wasn't even interested in the AI features, just wanted to try a fancy new terminal. Not saying warp might not be useful for some people, just wasn't for me. So far I've found explicitly calling for assistance with a CLI command (I've used aichat for this, but there's several out there) to be more useful in those occasional instances where I can't remember some obscure flag combination.
dimatura··on Tabular Programming: A New Paradigm for Expressive Computing
Didn't expect to see the M8 when opening this. It's truly an amazing device, and the tables are a really awesome addition to the typical audio modulation option.
dimatura··on Tabular Programming: A New Paradigm for Expressive Computing
that would be pretty confusing, as it happens M8 was partially inspired by another tracker called LGPT (Little Game Park Tracker, named after the device it ran on).
dimatura··on Variable duty cycle square waves with the Web Audio API
I'm not an expert on GB chiptune, but from what I've heard from enthusiasts is that different GB models sound different, and even within the same model there are variations. That said, it wouldn't surprise me if the GB waveforms aliasing, at least from the digital side, given that it was operating with pretty minimal synthesis capabilities. There's probably some extra low- and high-pass filtering that shape the sound after the waveform itself is generated. Looking at some examples of actual GB PWM waveforms, for sure some high-pass would make a pure PWM waveform more GB-like. And some low-pass could help a bit with aliasing.
dimatura··on Variable duty cycle square waves with the Web Audio API
Yeah, at least in the context of music synthesis, everyone just says PWM. (The term even being subject of a meme with a certain youtuber who seems to be fond of it ;)).
dimatura··on Baby Steps into Genetic Programming
Note that genetic programming is a specific subset of genetic algorithms that focuses on searching for "programs" (hence the name), typically encoded in the form of a tree structure similar to an AST, though other representations exist. In theory, you could use GP for almost any situation where you'd want to synthesize a mathematical expression or piece of code for which you have a criteria to optimize (i.e. "fitness"). In practice, since GPs are basically just semi-randomly searching a huge combinatorial space, they tend to work best in low-dimensional problems, ideally with a fitness function that is cheap to evaluate. They can work nicely for finding nonlinear symbolic formulas for regression, for example. But there's also some other cool results over the years - Hod Lipson has published several cool results in robotics with them.

Until a few years ago, the popular deep learning methods like CNNs weren't great at that kind of thing, but LLMs definitely changed that - it's safe to say that by drawing on huge amounts of data LLMs are much better at most practical programming tasks. They're not necessarily a complete replacement though, there's definitely space for hybrid approaches, eg https://arxiv.org/abs/2401.07102 or https://www.nature.com/articles/s41586-023-06924-6.

dimatura··on (2016) Interactive Neural Network Art
Yeah, if you look at the original NERF paper (which is not by Nvidia btw) they cite CPPNs as prior work.
dimatura··on NotaGen: Symbolic Music Generation
The pop composition was nice, if somewhat cliched. (And of course the MIDI instrumentation does it no favors). The guitar solo in the middle was fun - it kind of sounded like it reverted to some of the baroque pieces in its pretraining. Reminded me of the harpischord solo in the Beatles' "In my life".
dimatura··on Introduction to Deep Learning (CMU)
I took this course the first semester it was given. There was one TA, and now there's 24! Fun fact: The TA was the writer of Aqua's 90s hit "Doctor Jones".
dimatura··on Arbitrary-Scale Super-Resolution with Neural Heat Fields
Yes, there are plenty of them. Not sure what the SOTA is, though. Similar to super-resolution, it is relatively simple to create a nearly-infinite dataset for these; pick a clean image, then introduce JPG artifacts. Then train a model to invert the process.
dimatura··on Show HN: VSC – An open source 3D Rendering Engine in C++
povray is a raytracer, this is a rasterizer (like what openGL does, say). Different rendering techniques.
dimatura··on The Future Is Niri
A nice side benefit is that ratpoison is its own screen lock (for 99% of people).
dimatura··on The Future Is Niri
I still use ratpoison, dreading the day it stops working for whatever reason (wayland being a likely one). Haven't tried it, but Cagebreak (https://github.com/project-repo/cagebreak) seems like a possible successor.
dimatura··on DiffRhythm: Fast End-to-End Full-Length Song Generation with Latent Diffusion
I definitely see this happening. Music generation has lagged behind image generation but is following more or less the same path. Early image generation models were completely unconditional; all you could do was sample an image. Then coarse conditioning methods such as text prompts and depth images came along; then additional tooling to tune images in a more fine-grained way.

That said, there is a difference to images in that music also has a "symbolic" level to it that is closer to text than images [1]. There's other work out there that uses LLM-type tools for direct melody generation (no audio). And of course, there's lyrics. I do expect commercial tools to start integrating all these capabilities gradually, it's just a matter of time.

[1] I guess there's also vector images (like SVG) - I've seen work in generating those as well, though it's less mature than directly generating pixels.

dimatura··on DiffRhythm: Fast End-to-End Full-Length Song Generation with Latent Diffusion
One thing that strikes me about almost every AI-generated track (from academic or commercial generators), is that even if it's often "competent" - in that it has reasonable melodies, chord progressions, etc - is how average it is. Mediocre, taking the term literally. In a way that also highlights cliches and crutches that are common in human-made music. Somewhat reminiscent of GPT text that drones on and on in a grammatically correct way but conveys little of interest. This is of course not unexpected, given how these models are trained. I wonder if this will have an effect of pushing (human) musicians to be more experimental - to move away from the conventions that are now just a click away for anyone.
dimatura··on Tell HN: Generational Gap
I've noticed a lack of familiarity with design patterns in junior employees as well, but I've attributed it to them not paying attention in class and/or forgetting about it. Are design patterns not part of the standard curriculum these days? I guess to some degree, the most successful patterns are already "baked in" into frameworks and even languages, so maybe they're not even worth mentioning as something noteworthy?
dimatura··on A possible future of Python in the browser
As someone that mainly codes in Python and only has rudimentary JS skills, I'm intrigued. The generative art pyscript demos mentioned in the blog post are really nice.

Using decorator calls to "compile" chunks of python code seems like an increasingly popular trend in ML (and presumably other numerically intensive code) - e.g., numba, jax.jit and triton.

dimatura··on I think Yann Lecun was right about LLMs (but perhaps only by accident)
A more apt analogy would be a human trying to walk somewhere with their eyes closed, i.e., what you may know as open-loop control.
dimatura··on I think Yann Lecun was right about LLMs (but perhaps only by accident)
No, using a large context window (which includes previous tokens) is a critical ingredient for the success of modern LLMs. Which is why you will often see the window size mentioned in discussions of newly released LLMs.
dimatura··on I think Yann Lecun was right about LLMs (but perhaps only by accident)
I don't think whether LLMs use only the last token, or all past tokens, affects LeCun's argument. LLMs already used large context windows when LeCun made this argument. On the other hand, allowing backtracking does. Which is not something the standard LLM did back when LeCun made his argument.
dimatura··on Grok3 Launch [video]
I see what you did their
dimatura··on Benchmarking vision-language models on OCR in dynamic video environments
Re: reviewers, I don't see any mention of this being accepted into a peer-reviewed venue. Peer review isn't necessary for arxiv submissions.
dimatura··on Our channel on YouTube has been deleted due to “spam and deceptive policies”
That's shocking, sinevibes is a legit developer in the music/audio space. They made the FX for dreadbox's Typhon synth and several popular FX plugins for Korg logue synths. I'm subscribed to their newsletter and have never received spam. All their youtube channel had were demos of their products, as far as I recall. Like "here's how the dry synth sounds, here's how it sounds with reverb", can't even imagine how an algorithm thought it runs afoul of a spam or scam policy. Probably a mistake, hope it's fixed.
dimatura··on Show HN: ESP32 RC Cars
I put a link above for this kit which is about $50. However, I believe you would also need an ESC for the motor, a suitable one would probably be less than $20.
dimatura··on Show HN: ESP32 RC Cars
I believe it's this kit: https://www.elecrow.com/4wd-smart-car-robot-chassis-for-ardu.... It's been around a while, I got one around the time covid started. It's decent quality, although IIRC the front wheels don't have bearings.
dimatura··on A mouseless tale: trying for a keyboard-driven desktop
There's dozens of us!
dimatura··on No Calls
I'd wait to see if they have a good black friday sale.
← PreviousPage 3 of 15Next →