lostmsu

14 ms·

271.
by lostmsu 4mo ago
This article says nothing about the relative volume. Coming from the hands of OpenAI I'd consider it a form of peddling.
272.
by lostmsu 4mo ago
Just tried ping -s 1400 over Yggdrasil (IPv6 overlay) and it works. Reinforces my overlay choice. Getting screwed by browsers though because their WebRTC implementations completely ignore Yggdrasil addresses.
273.
by lostmsu 4mo ago
> (2-bit quantization for experts) with limited degradation. Certainly competitive with a 27B model Uh-huh...
274.
by lostmsu 4mo ago
27B does, the op was talking about consumer use of DS v4 Flash, that's 160GB.
275.
by lostmsu 4mo ago
> the choice of experts is expected to be highly biased Why? Why do you think that's the case? Part of the training is balancing load between experts. > so in aggregate there's a speedup. Yes. 2x. Over theoretical under 1 to
276.
by lostmsu 4mo ago
Before batching. The slowdown is because the model does not fit the xRAM so experts will have to be read from SSD on every forward pass. That's why it is impractically slow. Batching could allow you to generate 10 tokens for 10 differe
277.
by lostmsu 4mo ago
Why would he want it back?
278.
by lostmsu 4mo ago
They say they are using https://github.com/tile-ai/TileRT - persistent CUDA kernel - tiled processing with overlapping read/writes - model designed with specific constraints in mind
279.
by lostmsu 4mo ago
Cerebras currently does not provide any discounts for prefix caching making its use for agentic workloads sqr(n_turns) more expensive.
280.
by lostmsu 4mo ago
But they could be mingled with port numbers
281.
by lostmsu 4mo ago
Assuming you magically use all 128GiB of xRAM you need to read ~32GiB per token in batched mode. On a good SSD that would be 1/3 tokens per second. Cool, 2x that you can do 2/3 tokens per second. Let's assume you are lucky an
282.
by lostmsu 4mo ago
Yeah, that's what talking out of your butt is literally. "theoretical", no ballparks, ignorant assumptions about expert reuse.
283.
by lostmsu 4mo ago
Sounds like you're talking out of your butt instead of doing the math.
284.
by lostmsu 4mo ago
You can't batch MoE
285.
by lostmsu 4mo ago
DeepSeek V4 Flash is 160GB while Qwen 27B is about 27GB. You can't even run DS Flash on consumer platforms, let alone batch it.
286.
by lostmsu 4mo ago
They "address" it by making false statement that the video stream is highly predictable. Sure, you might be able to predict 99% of video stream (for which you'd need to have a physics model, negating the whole point of baby f
287.
by lostmsu 4mo ago
GPT-2 (2019) https://openai.com/index/better-language-models/ Forever reinforced by Humans Who Are Not Concentrating Are Not General Intelligences: https://srconstantin.wordpress.com/2019/02&#
288.
by lostmsu 4mo ago
Build a space exploration company?
289.
by lostmsu 4mo ago
Do you suppose people in occupied full room that's being irradiated must cover their faces? Read better.
290.
by lostmsu 4mo ago
Humor is not a replacement for an argument. Your joke is equivalent to "as long as it isn't a autism-inducing vaccine", but from the left.
291.
by lostmsu 4mo ago
I'd bet it's not my Minimax instead.
292.
by lostmsu 4mo ago
I don't believe it is feasible to run Minimax 2.5 on a 3k computer.
293.
by lostmsu 4mo ago
That depends on your dating goal.
294.
by lostmsu 4mo ago
Specific actions taken?
295.
by lostmsu 4mo ago
What do you know, certain forms of UV are used to disinfect skin. Too bad the developers of the concept didn't care at the moment they are "bonkers". https://www.canada.ca/en/public-health/services&#
296.
by lostmsu 4mo ago
Why do I have to present evidence for something I'm not claiming? Your response fits the picture drawn by the comment on out of context. > As for me, I don't think you need a phd to read the warning stickers on bleach and isopr
297.
by lostmsu 4mo ago
Are you asking me to find evidence counter to my understanding of situation? Show that he was prosecuted for copying. From a quick skim I'm in doubt anything he was found guilty of is copying.
298.
by lostmsu 4mo ago
Again, no quotes. But even as-is the idea of using ultraviolet to disinfect internal surfaces is not bonkers to any degree. Like I could see it working for nasal disinfection. In fact that's so obvious, that given you get very sad that
299.
by lostmsu 4mo ago
> being exposed to a "very powerful light" Isn't ultraviolet sanitization a thing? Are you sure this quote isn't another bit taken out of context that would further prove the point?
300.
by lostmsu 4mo ago
That man did not go to jail for any random copying.

More ›