petu

6 ms·

91.
by petu 5mo ago
> I can imagine it being ten or hundred times more I don't think I average even 2 captchas a day being terminally online, so 10 across every soul in the world sounds way too much for me. (we're ignoring bots it's meant to
92.
by petu 5mo ago
> but unless you do PoW (which is also ecologically a nightmare) Can you expand? I don't see a problem with some napkin math. 5W load for 2 seconds is 0.002Wh (we have to let smartphones pass and not by doing PoW for 10s of seconds)
93.
by petu 5mo ago
Owned by Lenovo, yes
94.
by petu 5mo ago
Note that Xiaomi today is very hostile to bootloader unlocking -- to get unlock code you need to win in daily first come, first served "lottery" (they limit total number of unlocked devices per day). After a second or two passes y
95.
by petu 5mo ago
Half of the world lives on $300/mo. For majority of the world there's meaningful impact in saving $20 on a book.
96.
by petu 5mo ago
It's digital copies, no real damage is done. AA asks you to not scrape them because of server load and provides torrents to download everything in more efficient manner.
97.
by petu 5mo ago
How do you know that Anna's archive started operating in 2022? edit: you've sent me Wikipedia link and then removed your reply. So I'll put my reply here: https://en.wikipedia.org/wiki/Anna%27s_Archive V
98.
by petu 5mo ago
formal Russian is 'em dash heavy', here's South Korea Wikipedia arcticle in Russian, Dec 2021 version: https://ru.wikipedia.org/w/index.php?title=%D0%A0%D0%B5%D1%8... Ignoring references, just in article
99.
by petu 5mo ago
> But looking closer, I actually suspect it’s not AI, the author just integrated LLM-isms into their style. I think his native/first language is Russian -- em dashes are widely used (e.g. most definitions start with it, look at any
100.
by petu 5mo ago
That's not SMTP or other industry standard protocol, incompatible with any mail client out there (?) -- it's for sending mail from backend services, not using as personal email. Even if you make it work with some SMTP adapter, you
101.
by petu 5mo ago
But it's only for receiving and you still need external mail server/address?
102.
by petu 6mo ago
> What makes the guess "right"? Matching token that would've been picked without speculative decoding. That seems to be more or less agreed upon. e.g. vLLM docs list tests they run to ensure that output doesn't change
103.
by petu 6mo ago
Speculative decoding batches multiple completions on all possible outcomes (0/1/2 draft tokens accepted) and sees if big model deviates at any point -- thus verifying each token. So there's no difference in output.
104.
by petu 6mo ago
Render tiles locally on iOS companion app side
105.
by petu 6mo ago
System checking it just verifies the signature is valid and thus all data presented is valid? Your browser doesn't need to query any Root CAs to trust SSL certificate, https works without internet. History of entry and visas/etc c
106.
by petu 6mo ago
> Relying upon the Internet being there for ID purposes is a massive fail. Why would you need internet? Document holder smartphone can cache the document for years and present it over NFC (including photo, signature, etc). Just like exis
107.
by petu 6mo ago
Models are math functions that predict next word, not conscious beings. If it was trained on dataset including data up to Q1 2025, then that's more or less expected answer -- even Qwen 3 didn't exist. If you see model that can rel
108.
by petu 6mo ago
> Qwen3.5-27b 8-bit quant 20 to 25 tok/sec It that with some kind of speculative decoding? Or total throughput for parallel requests?
109.
by petu 6mo ago
If they're two generations apart, then 200g difference could be hiding anywhere, but touchscreen (I really doubt it adds even 20g). e.g. 30g in larger heatsinks, 80g in glass-laminated screen, 20g in larger battery, 40g in more stiffer
110.
by petu 6mo ago
> Touchscreen does add about 200g of weight. 200g is weight of a smartphone, there's no way touch weighs that much. Framework 13 Pro screen seems to have plastic surface as before, not glass-laminated (which I guess could add 200g,
111.
by petu 6mo ago
Nobody said that. MacBooks of that period made compromises for useless gain in thinness. You can't with straight face tell that butterfly mechanism was a good tradeoff for .3 mm.
112.
by petu 6mo ago
> 155w max (but realistically more like 80w sustained) 155W PSU seems to be unified with M4 Pro model, plus there's reserve for peripherals (~55W for 5 USB/Thunderbolt ports). Apple lists 65W for base M4 Mac itself: https:&#x
113.
by petu 6mo ago
Google Maps app sees that you took photo near POI and later in the day asks you in notification if you want to share it on maps. You review the photo and go "lol, sure". At least for me that doesn't even feel like posting due
114.
by petu 7mo ago
It's not giving you access to Android shell, but one inside VM. So OS image for VM.
115.
by petu 7mo ago
> OpenAI released one once they've released gpt-oss-safeguard in October I hope / think they are going to release more, just going for one big release a year like Gemma (if we talk strictly about general chat model -- Gemma 3 w
116.
by petu 7mo ago
If you don't care about how it's architectured, why you care about size? Compare it to Q3.5 397B-A17B. Just like smaller size models are speed / cost optimization, so is MoE. G4 26B-A4B goes 150 t/s on 4090/5090, 80
117.
by petu 7mo ago
It means that model was tuned to to act as chat bot. So write a reply on behalf of assistant and stop generating (by inserting special "end of turn" token to signal inference engine to stop generation). Base model (without instruc
118.
by petu 7mo ago
Qwen 3.5 27B is dense, so (I think) should be compared to Gemma 4 31B. Or Gemma-4 26B(-A4B) should be compared to Qwen 3.5 35B(-A3B)
119.
by petu 7mo ago
Try 26B first. 31B seems to have very heavy KV cache (maybe bugged in llama.cpp at the moment; 16K takes up 4.9GB). edit: 31B cache is not bugged, there's static SWA cost of 3.6GB.. so IQ4_XS at 15.2GB seems like reasonable pair, but e
120.
by petu 7mo ago
If controller isn't being rotated in the moment, then camera should be stationary regardless of controller orientation, yea. I guess you've used some strange gyro to stick emulation (never heard of such thing, but sounds like it).

More ›