Y
HN Search
Hacker News Search
new
|
comments
|
top
|
jobs
spoaceman7777
searching PlanetScale…
1.
▲
2.
▲
3.
▲
4.
▲
5.
▲
6.
▲
10 ms
·
61.
▲
by
spoaceman7777
1y ago
All you really need to see is the picture of the CPU with thermal paste only on one half. Thermal throttling is tuned to work when there is 1. a sufficient heatsink (theirs was significantly below requirements) and 2. it is installed correc
62.
▲
by
spoaceman7777
1y ago
Not really a lot it can do rapidly enough if there's only thermal paste on half the CPU. It sounds like the user likely did the opposite of the "offset seating" of the heatsink that Noctua recommended.
63.
▲
by
spoaceman7777
1y ago
yes, but it puts you in the incredibly small bucket of "users that has weird headers that don't mesh well", and makes using the rest of the (many) other fingerprinting techniques all the more accurate.
64.
▲
by
spoaceman7777
1y ago
samsung is a lot better with keeping products staffed at least
65.
▲
by
spoaceman7777
1y ago
The only downside is that all of these new features will be supported for about 3 weeks, and then rapidly turn into another Google abandoned strip mall. Fooled me once, shame on you, fooled me (hundreds of times), shame on me Bet the people
66.
▲
by
spoaceman7777
1y ago
This is a bot hoax. The only news here is that twitter still hasn't fixed its insane spam account problem
67.
▲
by
spoaceman7777
1y ago
no, that's just because claude has been having persistent brown-outs for a while now. Microsoft and Anthropic messed up by bailing out of full-tilt AI hyperscaling a fair bit. (i.e., Microsoft cancelled a bunch of AI datacenter project
68.
▲
by
spoaceman7777
1y ago
The reality is the opposite of this post. LLMs are great at rapidly creating rough drafts, and humans are best (when properly trained) at critiquing LLM results. So, LLMs are in fact better at blue-teaming, and humans are better at red-team
69.
▲
by
spoaceman7777
1y ago
And don't forget xAI, which has MechaHitler in its product DNA
70.
▲
by
spoaceman7777
1y ago
Seems this is likely due to ongoing work on FP8 support on nvidia/cutlass. From my reading, the alternative code path was likely added recently for testing by external contributors to the cutlass project, and other involved parties. (R
71.
▲
by
spoaceman7777
1y ago
This article is entirely wild medical speculation, with a sample size of one, from someone who claims to have given away all of this Lumina product, and only using a bit of "residue". They also mention using a narcotic supplement
72.
▲
by
spoaceman7777
1y ago
Depends whether you asked it to just write the code, or whether you asked it to evaluate the strategy, and write the code if nothing is ambiguous. My default prompt asks the model to provide three approaches to every request, and I pick the
73.
▲
by
spoaceman7777
1y ago
The anthropic team released a paper a couple of days ago which demonstrated a similar effect with Claude 3.5 and other models, where changing the system prompt to tell it that it was created by other orgs or people drastically altered its c
74.
▲
by
spoaceman7777
1y ago
This is part of the basis for Supabase offering their realtime service, and broadcast, rather than supporting native LISTEN/NOTIFY. The scaling issues are well known.
75.
▲
by
spoaceman7777
1y ago
It still hasn't been turned back on, and that repo is provided by xAI themselves, so you need to trust that they're being honest with the situation. The timing in relation to the Grok 4 launch is highly suspect. It seems much more
76.
▲
by
spoaceman7777
1y ago
As far as I know, neither of them have had a card unintentionally light on fire. The whole thing started with Derbauer going to bat for a cable from some 3rd party vendor that he'd admitted he'd already plugged in and out of vario
77.
▲
by
spoaceman7777
1y ago
I'm referring to the section header in this article. Youtubers are not a truly hegemonic group, but there's a set of ideas and narratives that pervade the group as a whole that different subsets buy into, and push, and that's
78.
▲
by
spoaceman7777
1y ago
The real issue here is actually harebrained youtubers stirring up drama for views. That's 80% of the problem. And their viewers (and readers, for that which makes it into print) eat it up. Idiots doing hardware installation, with zero
79.
▲
by
spoaceman7777
1y ago
Using structured generation (i.e., supplying a regex/json schema/etc.) for outputs of models and tools, in addition to doing sanity checking on the values returned in struct models sent/received from tools, you are able to pr
80.
▲
by
spoaceman7777
1y ago
If you replace the word "LLM" in your reply with "web browser", I think you'll see that the situation we're in with MCP servers isn't truly novel. There are lots of tools to handle the many, many programs
81.
▲
by
spoaceman7777
1y ago
Wow. The only people who could have decided to write and publish this article are the same people who did all of the other insane things in that article. It reads as though they're begging people not to bail them out. Like... in all of
82.
▲
by
spoaceman7777
1y ago
It's also worth noting that it's almost entirely native born Americans that are pushing back against nepotism. Extreme nepotism is still the norm (an expectation even) in most South and East Asian cultures. And it's quite rea
83.
▲
by
spoaceman7777
1y ago
Looks like that was a hoax.
84.
▲
by
spoaceman7777
1y ago
I'd assume that, in the context of LLM inference, "recent" generally refers to the Ampere generation and later of GPUs, when the demand for on board memory went through the roof (as, the first truly usable LLMs were trained o
85.
▲
by
spoaceman7777
2y ago
the majority of global internet users access the internet via a phone.
86.
▲
by
spoaceman7777
2y ago
The Kalshi market for this has had 2:1 odds against them even reducing government spending, let alone lower the annual deficit (and lessening the debt is right out). The last Republican president that managed to cut annual spending was Eise
87.
▲
by
spoaceman7777
2y ago
> Adobe isn't trying to be ethical, they are trying to be more legally compliant Is the implication of this statement that using AI for image editing and creation is inherently unethical? Is that really how people feel?
88.
▲
by
spoaceman7777
2y ago
Didn't yet . Which is how most of this technology starts out, and is the largest problem with people building these systems. See: courts ruling that 23andMe can sell their data yesterday
89.
▲
by
spoaceman7777
2y ago
Exactly. I see these threads over and over, and it's just senior devs complaining about how it's the tool's fault, and not that they haven't put the time in to learn the new tool The cycle of tool and framework re-skilli
90.
▲
by
spoaceman7777
2y ago
I have in my prompt for it to always use search, no matter what, and I get pretty decent results. Of course, I also question most of its answers, forcing it to prove to me that its answer is correct. Just takes some prompt tweaking, redos,
More ›