Tiberium

13 ms·

271.
by Tiberium 9mo ago
I don't know if you can consider Netcup "small", but their RS 4000 G12 ("root server", basically a VPS with dedicated/guaranteed resources) costs ~€31 for a monthly contract for any location in Europe without
272.
by Tiberium 9mo ago
If anything, it seems that the author used GPT 5.2 (-codex) in Codex, which is actually far more capable at such work than Opus 4.5 in Claude Code.
273.
by Tiberium 9mo ago
Coca-cola's "secret formula" is also just marketing.
274.
by Tiberium 9mo ago
The developers have fixed it soon after the article was posted, although they seem to disagree with the article's framing: https://github.com/screeps/screeps/issues/162#issuecomment-3...
275.
XSS –> RCE in Screeps, a programming game on Steam(outsidetheasylum.blog)
4 points by Tiberium 9mo ago|3 comments
276.
ADSL Works over Wet String(revk.uk)
7 points by Tiberium 9mo ago|1 comments
277.
by Tiberium 9mo ago
LLM comments are not allowed on HN, plus the author is being dishonest in another comment saying that the style is a "deliberate style choice" and not because it's vibecoded. Perhaps it's not immediately obvious, but an
278.
by Tiberium 9mo ago
New account, vibecoded demo (the style gives it away, plus the comments in source), author using LLM to write comments, yet this reached the top page.. HN is certainly interesting nowadays. I'm not against vibecoded stuff, but the auth
279.
by Tiberium 9mo ago
At least I'm not the only one who noticed. It's genuinely weird and unsettling how such AI-written blog posts nowadays get to the top of HN easily.
280.
by Tiberium 9mo ago
Great project! Just wanted to talk about the "/ The edit problem" section in the blog post - it's almost entirely false. Inpainting (when you provide a mask) has existed for years , and for good edit models (such as Nan
281.
by Tiberium 9mo ago
The project (even if it's made with the help of LLMs) is nice, but the author writing all of his HN comments with LLMs is not.
282.
by Tiberium 9mo ago
I find such "X in Y lines of code" challenges not very interesting most of the time, because, as it is the case here, they usually just pack multiple lines into one instead of using clever tricks, from one of the lines in that fil
283.
by Tiberium 9mo ago
At this point it feels like HN is becoming more like Reddit, most people upvote before actually checking the repo.
284.
by Tiberium 9mo ago
Usually they are smaller details in the pages, not the core claims, but that doesn't really refute my point of GPT being easily able to point them. Here are some examples, I'm not including all errors per page that GPT 5.1 found
285.
by Tiberium 9mo ago
It gives references that you can then verify manually. I wasn't advocating for a 100% automated process.
286.
by Tiberium 9mo ago
Yes, I did.
287.
by Tiberium 9mo ago
You can easily do this with normal GPT 5.2 in ChatGPT, just turn on thinking (better if extended) and web search, point a Wikipedia page to the model and tell it to check the claims for errors. I've tried it before and surprisingly it
288.
by Tiberium 9mo ago
Small bug: the sim seems to double count residential population for taxes, industrial pop gets ignored. The duplicate logic seems to exist in UI but it's correct there. Also I wonder if all the repetition could be reduced, or would it
289.
by Tiberium 9mo ago
A bit unrelated, but if you ever find a malicious use of Anthropic APIs like that, you can just upload the key to a GitHub Gist or a public repo - Anthropic is a GitHub scanning partner, so the key will be revoked almost instantly (you can
290.
by Tiberium 9mo ago
Was a bit disappointed when I almost "solved" it but couldn't solve the last 2 words, finally clicked the hint and it told me to undo 12 times.. would have preferred if there was a warning earlier.
291.
by Tiberium 9mo ago
Can anyone point to an actual reputable source that has any details about what specifically got leaked, and how? Instagram has way more users, so it's very odd that only 17.5M get "leaked". Just honestly feels like this is ov
292.
by Tiberium 9mo ago
Is there a reason why you're using LLMs for comments as well?
293.
Constitutional Classifiers++: More efficient protection against jailbreaks(anthropic.com)
3 points by Tiberium 9mo ago|0 comments
294.
by Tiberium 9mo ago
No, for example Alibaba has huge proprietary Qwen models, like Qwen 3 Max. You just never hear about them because that space in western LLM discussions is occupied with the US labs.
295.
Kahvibreak – J2ME game collection (5600+ games)(bluemaxima.org)
3 points by Tiberium 10mo ago|0 comments
296.
by Tiberium 10mo ago
Oh, of course not, you might need up to 100GB VRAM to have those models at decent speeds even just for low-quant versions. And all the hype about Macs with unified memory is a bit dishonest because the actual generation speed will be very b
297.
by Tiberium 10mo ago
The best is probably something like GLM 4.7/Minimax M2.1, and those are probably at most Sonnet 4 level, which is behind Opus 4.1, which is behind Sonnet 4.5, which is behind Opus 4.5 ;) And honestly Opus 4.5 is a visible step change a
298.
by Tiberium 10mo ago
I think the title should clarify the year - (2024), because those tools are not useful in the way artists want them to be.
299.
Hardcoded Auth Token in RustFS (CVE-2025-68926)(github.com)
5 points by Tiberium 10mo ago|1 comments
300.
by Tiberium 10mo ago
Yeah, it's surprising that people just blindly believe posts like that.

More ›