HackerTrans
TopNewTrendsCommentsPastAskShowJobs

amarshall

no profile record

comments

amarshall
·24 dni temu·discuss
Pretty light on details, heavy on fluff. 9.8s to 3.1s was userfaultd + hugepages, 500ms was PS/2 mouse and… where is the rest of the time to get to 400ms?
amarshall
·24 dni temu·discuss
No, and having that is not required to know output speed nor the effect of thinking, so I don’t see the point in such a superfluous, indirect question.

As for the question you’re likely asking: benchmarks that include speed across many models and providers available at various places e.g. https://artificialanalysis.ai/leaderboards/models
amarshall
·26 dni temu·discuss
Thinking doesn’t change output speed. Anthropic’s models are ~ 40–60 t/s median output speed.
amarshall
·26 dni temu·discuss
Can take the geometric mean of total and active parameters of MoE to get approximate equivalent quality to dense model params. So sqrt(35*10)≈18.7.

The trade-off of MoE is that it is worse but faster for the same total size.
amarshall
·26 dni temu·discuss
What context length and kv cache quant (if any) are you using? And MTP?
amarshall
·2 miesiące temu·discuss
Probably want `kitty --single-instance` to reduce memory usage when opening multiple.

https://sw.kovidgoyal.net/kitty/invocation/#cmdoption-kitty-...
amarshall
·2 miesiące temu·discuss
The actual paper (linked in another comment here) is literally titled “Nonergodicity and Simpson’s paradox in neurocognitive dynamics of cognitive control”. Why they omit it from the blog post version, I’ve no idea.
amarshall
·2 miesiące temu·discuss
Seems like a case of Simpson’s Paradox https://en.wikipedia.org/wiki/Simpson%27s_paradox
amarshall
·3 miesiące temu·discuss
If you are running on 4090 and get 5 t/s, then you exceeded your VRAM and are offloading to the CPU (or there is some other serious perf. issue)
amarshall
·4 miesiące temu·discuss
Oh I missed the "quad" before 3090.
amarshall
·4 miesiące temu·discuss
You're almost certainly (definitely, in fact) confusing the 120b and 20b models.
amarshall
·4 miesiące temu·discuss
Yes. Even 90 Hz is a noticeable improvement over 60 Hz. I wouldn’t pick it over high-DPI, though.
amarshall
·6 miesięcy temu·discuss
No, the connector is longer than it is tall.
amarshall
·6 miesięcy temu·discuss
Maybe, but Apple doesn’t make them thinner anyway so the argument is invalid. iPhone 6S with headphone jack: 7.1mm thick. iPhone 17 is 7.95mm thick.
amarshall
·7 miesięcy temu·discuss
It’s easier than that. Replace shorts with watch in the URL.
amarshall
·8 miesięcy temu·discuss
Non-Shorts can be vertical as well. It works fine, and without the swipe-encouraging UX. If they were “just making long/medium-form videos in portrait now” they don’t need YT Shorts to do that.
amarshall
·8 miesięcy temu·discuss
DisplayPort 2.1 (which the monitor supports) provides sufficient bandwidth for 7680x4320@60 Hz 10-bit without DSC when using UHBR20. The press release unfortunately doesn’t clarify whether the monitor supports UHBR20 or only the lower UHBR10 or UHBR13.5 speeds. Of course, the GPU must also support that (Nvidia RTX 5000 only at the moment, as I believe AMD RX 9000 is only UHBR13.5).
amarshall
·10 miesięcy temu·discuss
Redownloading everything isn’t a risk when the lock file contains a hash of the download on first update.
amarshall
·10 miesięcy temu·discuss
Enabling Reduce Transparency is a sure-fire way to find a dozen bugs within a few hours. It is always quite apparent no one who is empowered tests it at Apple. At least the padding issue is the one Feedback report I sent that got the “more than 10 similar issues” label so it may actual get fixed.
amarshall
·10 miesięcy temu·discuss
> SO much padding

No idea on macOS, but turn on Reduce Transparency on iOS and there’s tons of padding most of the time, but then sometimes zero padding. And I mean zero. The edges of buttons and text are at the edge of the underlying background. It’s…embarrassing.