HNHacker News
TopNewBestAskShowJobs

machiaweliczny

695 karma · joined July 12, 2015

submissionscomments
machiaweliczny··on I'm switching to Python and actually liking it
I use "tsx script.ts" often and it's very nice good due to gradual typing you can use (plus I know lang well)
machiaweliczny··on I'm switching to Python and actually liking it
Agree that celery is a mess and it doesn't work well with async (Asyncio) python. I think version 6 maybe will support it sometime.

I also had a lot of problem due to async primitives with sqlalchemy - there's some tricky stuff with asyncio.gather vs TaskGroup and how sqlalchemy session works with it to be able to compose code easily.

machiaweliczny··on I don't think AGI is right around the corner
Good point
machiaweliczny··on I don't think AGI is right around the corner
My layman take on it:

1) We need some way of reliable world model building from LLM interface

2) RL/search is real intelligence but needs viable heuristic (fitness fn) or signal - how to obtain this at scale is biggest question -> they (rich fools) will try some dystopian shit to achieve it - I hope people will resist

3) Ways to get this signal: human feedback (viable economic activity), testing against internal DB (via probabilistic models - I suspect human brain works this way), simulation -> though/expensive for real world tasks but some improvements are there, see robotics improvements

4) Video/Youtube is next big frontier but currently computationally prohibitive

5) Next frontier possibly is this metaverse thing or what Nvidia tries with physics simulations

I also wonder how human brain is able to learn rigorous logic/proofs. I remember how hard it was to adapt to this kind of thinking so I don't think it's default mode. We need a way to simulate this in computer to have any hope of progressing forward. And not via trick like LLM + math solver but some fundamental algorithmic advances.

machiaweliczny··on Triangle splatting: radiance fields represented by triangles
AFAIK Gaussian Splatting is somehow connected to NeRFs (neural radiance fields), so job of turning multiple 2D images into 3D scene. Actually tried doing something like this recently for drone navigation (using older point cloud methods) but no luck so far.

Can anyone who read this suggest something to use to scan room geometry using camera only in real-time (with access to beefy NVIDIA computer if needed) for drone navigation purposes?

machiaweliczny··on Claude 4
I personally use GPT 4.1 in simple ask mode most recently. Fast and usually correct for quite complex function so OpenAI seems to be winning IMO.

All these "agentic" things make these models so confused that it almost never gives good results in my testing.

machiaweliczny··on Gemini Diffusion
I feel like diffusion would be much more useful for code it it could only mark tokens as "valid" if they were passing code checks. So it could be thought as adding more of "semantic chunks" instead of just words. Not sure how to validate it as some additions always will result in invalid code. You could argue that running tests, linters is the same but I think one could make it that generations are validated much more often with diffusion models.

Example: You remove some function, you also remove all uses of it. You can't use not existing variable etc. This could be trained on well commited git repos or stalking/stealing the work of developers via editor

machiaweliczny··on Solidjs: Simple and performant reactivity for building user interfaces
Is there equvalent to SvelteKit for Solidjs?
machiaweliczny··on Why is OpenAI buying Windsurf?
IMO people underestimate effort of hiring people with domain expertise for this. Maybe they can try to buy one person from these teams and then have it help build the core IDE team but this will take much longer than just buying a very strong team.
machiaweliczny··on Zod v4 Beta
Were union types fixed? I remember that previously they didn’t work too well.
machiaweliczny··on Recent AI model progress feels mostly like bullshit
Because it’s lossy compression. I also consumed a lot of books and even more movies and I don’t have good memory of it all. But some core facts and intuition from it.
machiaweliczny··on Pierogi in Space
It feels like name pressure to become astronaut.

His name could be loosely translated as „Glorious Achievement-Cherry”.

machiaweliczny··on Why I find diffusion models interesting?
I actually think that diffusion LLMs will be best for code generation
machiaweliczny··on Using GRPO to Beat o1, o3-mini and R1 at "Temporal Clue"
Would be great if some details given about how exactly model is penalized for staying off-track.
machiaweliczny··on Apple's Software Quality Crisis
What about “Pop! OS” ?
machiaweliczny··on The top 10% owns 87% of the stocks
Problem Is government as they should know exactly what is wrong - that is their job
machiaweliczny··on Zelensky leaves White House after angry meeting
You seem to be missing $15T in minerals and for sure we can’t afford a freeze on current lines and this will make Russia invasion very profitable.
machiaweliczny··on Svelte 5 is not JavaScript
What most people need is good routing solution but it’s hard to make it with SSR. Seems like frameworks are somewhat converging now on this app router concept.

Agree that JSX is great (but styles like in Svelte also).

machiaweliczny··on Svelte 5 is not JavaScript
Have you used MobX? It was there 10 years ago and still working. Svelte runes are basically the same with some perf wins but require compilation
machiaweliczny··on The U.S. needs a shipbuilding revolution
True, Hitler mostly lost a war due to megalomania and spending 1/4th of naval budget on big destroyer instead of producing more ubots.
machiaweliczny··on Open Euro LLM: Open LLMs for Transparent AI in Europe
EU has enough capital it just works on US market cause there no unified capital market in EU. If Trump hits EU with tariffs I hope this changes.
machiaweliczny··on Open Euro LLM: Open LLMs for Transparent AI in Europe
Do they have something tangible now? At least PWr (from Poland) have fine tuned Mistral and a lot of language data AFAIK - https://bielik.ai/

Anyway I think some novel training method is needed and I’m with Yann on this. They could luck into good idea but with this budget chances are < 1%

machiaweliczny··on Introducing deep research
This is very bearish for current AI. Seems like 99% reliability is still too small with compounding errors. But I wonder of this is inherently specific to longer context or if this just depends on how it’s trained. In theory longer context => more errors

Although I think people are the same, too big problem and you are getting lost unless taking it in bites, so seems like OpenAI implementation is just bad because o3 hallucination benchmark shouldn’t lead to such poor performance

machiaweliczny··on Bacteria (and Their Metabolites) and Depression
I had some gut bacteria issues and I think I was helped recently by using oregano oil and/or allicine (extract from garlic). Used both for a week, before had 2x antibiotics for suspected SIBO. I also used VivoMaxx 450 probiotics and sodium butyrate for 3 days so maybe it was it. All of these combined are cheaper than single curation with antibiotics so worth giving a shot in case of issues.
machiaweliczny··on I had to take down my course-swapping site or be expelled
IMO it should be ranked order and random assignment
machiaweliczny··on Stay Gold, America
I wish more people were aware of this
machiaweliczny··on Search ArXiv Fluidly
Embeddings are very cheap to generate
machiaweliczny··on Map of GitHub
Embeddings are super cheap to compute
machiaweliczny··on Gemini 2.0: our new AI model for the agentic era
Yeah, I liked my windows phone, not sure why they killed it
machiaweliczny··on OpenAI, Google and Anthropic are struggling to build more advanced AI
Long context is a scam. Claude is best but it’s still gets lost with longer context
← PreviousPage 4 of 29Next →