jug

11 ms·

151.
by jug 1y ago
Just posting to say everything is fine. It works to report this way too, via .NET maintainers. It is very common to do this as well. https://github.com/dotnet/runtime/issues/117233#issuecomment... You’re quot
152.
by jug 1y ago
The only problem is that the generated comics have typos and seem to make little sense.. They’re just kind of setting up a story? Edit: Oh, I missed the choose your adventure aspect to them, so what’s on the front page are just samples of w
153.
by jug 1y ago
This is the first time I’ve seen HDR used to refer to wider color spaces and not extended brightness and contrast ratios.
154.
by jug 1y ago
They have even renamed Office to Microsoft 365 Copilot. Yes. Microsoft Office. https://www.windowslatest.com/2025/01/18/microsoft-just-rena...
155.
by jug 1y ago
Yeah it’s really annoying how opaque they are about the model there. Always just ”GPT 4 based” or ”GPT 4o based” if you dig in their blog archives. Makes one unable to check it against benchmarks or see when it’s updated. Setting expectatio
156.
by jug 1y ago
And yet, there are some notable differences between them, so now that there’s a benchmark and attention given to this issue, I wonder how much better they can get. Because obviously something can be done.
157.
by jug 1y ago
An AI generated status page would be the epitome of 2025.
158.
by jug 1y ago
And unfortunately revealed to be largely a vibe check these days with that whole Llama 4 debacle. But why should we be surprised, really, when users have an easier time feeling if the replies sound human and conversational and _appear_ know
159.
by jug 1y ago
There is no real need and the issue is hypothetical?
160.
by jug 1y ago
Before that, you might ask ChatGPT to create a vector image of a pelican riding a bicycle and then running the output through a PNG to SVG converter... Result: https://www.dropbox.com/scl/fi/8b03yu5v58w0o5he1zayh&#
161.
by jug 1y ago
Ugh, now I want to pick up F# again. I like and identify with many points about Haskell but F# fits like a glove for our ecosystem here.
162.
by jug 1y ago
Until now with a far better telescope able to significantly improve the sample size, that is. Ugh, this is so frustrating. We know our current theories cannot be complete but the LHC has mostly just confirmed assumptions, and now this. Ever
163.
by jug 1y ago
> Leta aims to present a reliable and trustworthy way of searching privately on the internet. > Leta is also useless if your browser blocks all cookies, tracking pixels and other tracking technologies. Huh? This needed better clarific
164.
by jug 1y ago
Given the shady stuff there, sounds like a match made in heaven to be honest, haha...
165.
by jug 1y ago
This is my feeling too, across the board. Nowadays, benchmark wins seem to come from tuning, but then causing losses in other areas. o3, o4-mini also hallucinates more than o1 in SimpleQA, PersonQA. Synthetic data seems to cause higher hall
166.
by jug 1y ago
Why would you not link against the UCRT if you’re intentionally designing for Win32 though? It’s a decade old library by now. Time flies…
167.
by jug 1y ago
Ooh, it’s the ”softening the blow ahead of dramatic and expected to be highly controversial changes” writeup! I think many have seen the writing on the wall, and it’s too bad. We love AI as SO replacers today, as they have an unprecedented
168.
by jug 1y ago
I’ve seen benchs on hallucinations and OpenAI has typically performed worse than Google and Anthropic models. Sometimes significantly so. But it doesn’t seem like they have cared much. I’ve suspected that LLM performance is correlated to ri
169.
by jug 1y ago
Edge also does this, as in not just web rendering for those parts of Windows using WebView2, but Edge itself. You can supposedly disable it in Edge -> Settings -> System and Performance -> Disable Quick Start
170.
by jug 1y ago
We better not only use these to burn the last, flawed model, but try these again with the new. I have a hunch the new one won’t be very resilient either against ”positive vibe coercion” where you are excited and looking for validation in mo
171.
by jug 1y ago
Weird. Using a custom solid color background works in my install of macOS. But I do use 15.4.1 in case the fix was just released.
172.
by jug 1y ago
Yeah, it's a funny take because this is in fact a more advanced form of AI with autonomous tool use that is just now emerging in 2025. You might say "They could search the web in 2024 too" but that wasn't autonomous on i
173.
by jug 1y ago
There’s a Reddit thread that provides useful context to this, what it is and the scope: https://www.reddit.com/r/math/s/OD0Jy9Rdns
174.
by jug 1y ago
Red flag nowadays is when a blog post tries to judge whether AI is AGI. Because these goal posts are constantly moving and there is no agreed upon benchmark to meet. More often than not, they reason why exactly something is not AGI yet from
175.
by jug 1y ago
I think since Windows 7, it has been very solid except for maybe third party kernel driver issues which needs direct kernel access out of necessity. Today when you have a BSOD, it’s like a Linux, Mach kernel panic etc - 99% likely to be a h
176.
by jug 1y ago
True, probably pretty feature complete after all these years although a pretty major lacking feature for any serious use is no UNC path support.
177.
by jug 1y ago
From the examples, it looks like this command-line tool should be able to be generalized to catalogue arbitrary data (retro gaming, tabletop games, movies, ...) with the same graphs/statistics/table, only with slightly larger over
178.
by jug 1y ago
Yes, as late as earlier today, I asked it to provide "naive" code which helped a bit.
179.
by jug 1y ago
Yes, this one is addictive for its speed and I like how Google was clever and also offered it in a powerful reasoning edition. This helps offset deficiencies from being smaller while still being cheap. I also find it quite sufficient for my
180.
by jug 1y ago
AI as we know it (GPT based LLM’s) have peaked. OpenAI noticed this sometime autumn last year when would-be GPT-5 was unimpressive despite the huge size. I still think ChatGPT 4.5 was GPT-5, just rebranded to set expectations. Google Gemini

More ›