efficax

13 ms·

181.
by efficax 11mo ago
realistically what’s the point of all those pixels at 32 inches? 5k at 27 inches seems more than enough.
182.
by efficax 11mo ago
what we can say right now is that we've hit the point of diminishing returns and the only way we're going to get signicantly more capable models is through a technological advance that we cannot forsee (and that may not come for d
183.
by efficax 1y ago
i've never heard anyone say "earphones"
184.
by efficax 1y ago
The answer is always amphetamines
185.
by efficax 1y ago
Try ~300k for an 8xH100 lol
186.
by efficax 1y ago
A silicon valley startup would never say one thing and then do another!
187.
by efficax 1y ago
he's not a very smart man
188.
by efficax 1y ago
> Once they tap into that for PPC/paid ads, If they start showing ads based on your prompts, and your history of "chats", it will erode the already shaky trust that users have in the bots. "Hallucinations" are on
189.
by efficax 1y ago
i do know how hard it is, that doesn’t change anything. nobody is forced to use something like zed and if you think it’s important enough to stress about the product decisions they make, the open source license gives you freedom to decide w
190.
by efficax 1y ago
Zed is fully open source. Fork it. The code is pretty nice, too, easy to understand.
191.
by efficax 1y ago
even if this is true as described, it doesn’t matter much: llms will become commoditized in the next few years and the chinese will have just as much access to them as anyone else. there’s no moat
192.
by efficax 1y ago
you raise interest rates to slow economic activity, you lower them to speed it up. yes, if interest rates are raised, that means employment is too high, basically, and they want to lower it.
193.
by efficax 1y ago
interest rates are controlled by central bankers, not magic. they make decisions based on their analysis of the economy. they raised rates to slow down the rate of investment and to suppress wages, in order to get inflation under control. L
194.
by efficax 1y ago
trust what? gimme that 14 billion and i don’t care what you do
195.
by efficax 1y ago
what the super-rich can buy that the middle class can't isn't really things (except for luxury goods, which are meh), it's people (through services). you can buy someone to do anything that you want to be done (that can be do
196.
by efficax 1y ago
waymo is pretty amazing but it's still short of complete full self driving, and it's not at all clear that it can close the gap. similary, current LLMs are really remarkable. but it's not at all clear that they will make the
197.
by efficax 1y ago
https://dlang.org/articles/safed.html check out the safed subset of the D language
198.
by efficax 1y ago
Tor might be good karma, but running a relay node is also a good way to be complicit in the sharing of illicit materials. Run this at your own risk
199.
by efficax 2y ago
Neat idea, but the execution seems terrible. I tried to find my coworkers based on specific searches and they never show up.
200.
by efficax 2y ago
they also look like they're using studio displays, instead of the pro display xdr, so they must just be doing the cuts here and the HDR color grading somewhere else. maybe they rent an editing room with those $30k+ reference displays t
201.
by efficax 2y ago
it’s common to be imprecise and mix up the type and an instantiation of the type, and that’s all that is happening here
202.
by efficax 2y ago
I've never seen a marketing website gated like this. Queue the user for the purchase, I just want to see what the dang thing is
203.
by efficax 2y ago
haven’t measured but the 70b runs fine on an m1 macbook pro with 64gb of ram, although you can’t do much until it has finished
204.
by efficax 2y ago
Widespread adoption of python did not really take place before 1.4/1.6 in the late 90s or really 2.0 in 2000.
205.
by efficax 2y ago
life in prison was too harsh, but a full pardon is too lenient.
206.
by efficax 2y ago
https://pharo.org/success/ It's used for real application development. It's honestly a great environment for building software
207.
by efficax 2y ago
of course they are. it’s simple: if they worked they would be incorporated into the loss function of the models and then they would no longer work
208.
by efficax 2y ago
bubbles go up and up and up… until they don’t
209.
by efficax 2y ago
You hear this about refactoring a lot and I don't get it. I've done several multi-thousand line refactors of a 500k line rust codebase in the past few years, and it's absolutely fantastic for this task. The type system guid
210.
by efficax 2y ago
It's an iMac, the target market is not running local LLMs or doing machine learning research

More ›