Zambyte

6 ms·

91.
by Zambyte 1mo ago
I have no historical context (heh) for this person, but as someone who uses LLMs regularly, I feel like there are reasonable takeaways from the article. Namely: - Don't vibe you agents.md file, it won't capture any intuition about
92.
by Zambyte 1mo ago
I have in mine to not do any write operations with git, to only use the CLI to explore the history and the current state. I don't trust my agent to commit and push on my behalf. Using my agents.md has worked well for that kind of thing
93.
by Zambyte 1mo ago
To be against copyright is to not believe the idea of "adversarial copying". It doesn't matter if you like or dislike what other people do with their copies. The act of copying simply does not harm you. The only way people ca
94.
by Zambyte 1mo ago
Why use Grok?
95.
by Zambyte 1mo ago
Probably. You can solve it with either some grounding context, or spending hundreds or thousands a month extra on a model that has more knowledge baked in. With modern harnesses, the choices is obvious.
96.
by Zambyte 1mo ago
Drew convinced me that we should abolish copyright through his blog[1][2]. I find it very unfortunate that he uses the copyright licenses of the software hosted on SourHut as an argument against LLMs. Abolishing copyright means abolishing c
97.
by Zambyte 1mo ago
I think the temptation for most people is not to offload your idea generation to AI, but to offload your storage and/or retrieval processes to AI.
98.
by Zambyte 1mo ago
I never suggested people should abandon cars for AI, I think people should abandon their primary mode of transportation for the same arguments that people usually make against AI. Cars have the same negative effects times 10 per person wh
99.
by Zambyte 1mo ago
AI doesn't rub rubber on asphalt at 70 mph. Did you know that? Electric cars and ICE cars both use more watts than AI. If you're concern is energy consumption, cars are worse. If your concern is pollution, cars are worse. Do you c
100.
by Zambyte 1mo ago
It's not hard to understand. What's hard to understand is why when I bring up how cars are literally orders of magnitude larger polluters and how we need to eradicate personal car use, people suddenly stop caring about pollution,
101.
by Zambyte 1mo ago
You don’t get to 25 or 30 trillion total addressable market, full stop. It's just a straight up pipe dream. If everyones jobs are automated away, there won't be any customers for the products and services provided by the companies
102.
by Zambyte 1mo ago
> How is anti pollution argument pro ai? What? (Edit: I think you misread "in support of their primary concern, being AI" as "in support of AI"? People who complain about pollution are usually mostly concerned about A
103.
by Zambyte 1mo ago
Also running Qwen 3.8 27b, on a 7900 XTX. I've been able to comfortably do a large majority of my programming locally since about Qwen 3.5. I also use pi.dev as my agent harness, and ollama as my inference engine. Though I've been
104.
by Zambyte 1mo ago
Eh, a healthy web is a web. I enjoy my preferred search engine, but surfing the web is becoming a lost medium.
105.
by Zambyte 1mo ago
It seems like there will be a collapse in building data centers, to the point that a large amount of the newly built ones will be reclaimed for other things. There is a very large group of people who are against building new data centers be
106.
by Zambyte 1mo ago
For the sake of your mental health, please ride an e-bike.
107.
by Zambyte 1mo ago
And I use SOTA models that are hosted on hardware I don't have access to a handful of times a month at this point, and most of the people I talk to regarding LLM usage are either in the same boat, or are asking about hosting things the
108.
by Zambyte 1mo ago
TFA isn't about LLMs intelligently escaping their inference engine, it's about people crafting malicious LLMs that exploit vulnerabilities in things like the inference engines parser. This is exactly the same problem that was once
109.
by Zambyte 1mo ago
Seriously. People fuss over LLM energy consumption, but I did the math on my rig, and if I were to use my machine at full tilt for all of my waking hours, I would use roughly 1/10th the amount of energy the average American uses drivin
110.
by Zambyte 2mo ago
How is it resistant? Can't all the pdfs just be leaked?
111.
by Zambyte 2mo ago
Thanks, I did put a decent amount of time into that. I couldn't get any model to one shot a truly homoiconic C, it took some coaxing, even from SOTA models. Maybe I caught a length based flag or something too, since it was fairly long.
112.
by Zambyte 2mo ago
I guess this is what happens when people separate the art from the artist, they get millions in funding. For the uninformed: https://news.ycombinator.com/item?id=45454026
113.
by Zambyte 2mo ago
I think in the age of AI slop contribution spam, pi has a notably good way to manage contributions. https://github.com/earendil-works/pi/blob/main/CONTRIBUTING....
114.
by Zambyte 2mo ago
And then everyone would have a user level parser instead of a homoiconic variant of C.
115.
by Zambyte 2mo ago
LLMs have been able to make the leap between generating and sending for years now. I'm not sure what your point is.
116.
by Zambyte 2mo ago
Consistent latency is generally more important for gaming than low latency. If I had to pick between a baseline of 8ms but with enough spikes to > 200ms to bring the average to 30ms, or a consistent 90ms with no spikes or dips, I would c
117.
by Zambyte 2mo ago
Blackmail, incitement, libel, and basically anything that comes up if you search "computer crimes".
118.
by Zambyte 2mo ago
Some things I really like are: - Generally high quality boards. I expect to be using my ZSA boards for decades. - Uses open, standard, programmable firmware (qmk). - They have fostered a community around their boards (importantly making it
119.
by Zambyte 2mo ago
That does look very nice. My mentor at my first job was actually the one who got me interested in split keyboards. His was a Kineses Advantage 2, which is also concave.
120.
by Zambyte 2mo ago
No, that would not count as homoiconic. Homoiconicity means that a language natively supports its own syntax as a data structure. I used AI to help write an example of what C would look like if it were homoiconic. You should mainly look at

More ›