VHRanger

26 ms·

331.
by VHRanger 3y ago
Is there a reason there's comparatively little storage in your machines in relation to RAM or even CPUs? Do your machines do compute heavy loads or something? For a DB I'd expect a lot more storage per node
332.
by VHRanger 3y ago
Typical American thinking. Your model 3 can't handle a corner. The reason car enthusiasts like Porsches is that they handle particularly well.
333.
by VHRanger 3y ago
The problem with Ikea is once you have to move them around. Or if you have kids/dogs jumping up and down the sofa, etc. The particleboard connections aren't very sturdy.
334.
by VHRanger 3y ago
There's a bunch of stuff in the way when installing fiber. Also humans telling you you can't build there. Starlink also makes sense for rural uses.
335.
by VHRanger 3y ago
> I think people mainly think of it as "their home and their community" rather than "economic extraction" It doesn't matter what they think. Some people also talk of "protecting property values". At the
336.
by VHRanger 3y ago
New residents don't just include immigrants: it's also all future inhabitants of the place. Your perspective amounts to existing residents extracting all the economic value in a location simply because they bought land there first
337.
by VHRanger 3y ago
Nvidia's stock price isn't at an all time price because of all the people writing fluid dynamics in CUDA. Nor is it because of all the tensorflow models people are writing, to be honest.
338.
by VHRanger 3y ago
> FLOPs utilization is arguably the industry standard metric for efficiency right now I'd argue GB/s memory bandwidth is more worried about at the moment.
339.
by VHRanger 3y ago
That's just because they don't have competition. Back in the dark days of 2015 we used to spend a day or two just getting tensorflow working on a GPU because of all the install issues, driver issues, etc. Theano was no better, but
340.
by VHRanger 3y ago
Daily reminder that data isn't the multiple of anecdote.
341.
by VHRanger 3y ago
Fair enough. I think people skip over that the vectors are the result of the minimization of the objective. That objective is roughly the same since word2vec. GLoVe is mathematically equivalent. LLMs are also equivalent. For a LM, the objec
342.
by VHRanger 3y ago
Oh, thanks for the correction. If all the vectors are on the unit ball, then cosine = dot product. But then the dot product is a linear transformation away from the euclidean distance: https://math.stackexchange.com/question
343.
by VHRanger 3y ago
It's isomorphic when vectors are normalized, otherwise it's angle distance, not position distance
344.
by VHRanger 3y ago
The scale of word embeddings (eg. Distance from 0) is mainly measuring how common the word is in the training corpus. This is a feature of almost all training objectives since word2vec (though some normalize the vectors). Uncommon words hav
345.
by VHRanger 3y ago
Brand?
346.
by VHRanger 3y ago
docs.tenstorrent.com All the details you need!
347.
by VHRanger 3y ago
> Nvidia has insane margins so it isn't hard* for a well-funded player to beat them in FLOPS/$ It's hard when you can't buy TSMC capacity
348.
by VHRanger 3y ago
This card isn't competing on price - it's a dev kit.
349.
by VHRanger 3y ago
Not really, Xeon Phis were basically the a x86 CPU abusing AVX extensions to their limit
350.
by VHRanger 3y ago
And why do they specify PCIe 4.0? I imagine it's just to leverage bandwidth correctly? Or is there a necessary feature in there?
351.
by VHRanger 3y ago
"The regime" You probably consume too much right wing nonsense. I'd check my media diet were I you.
352.
by VHRanger 3y ago
> I wonder why they are starting with these models. They were building Greyskull in 2020/2021 initially. BERT_Large and similar were the SOTA at that time.
353.
by VHRanger 3y ago
Much more efficient per watt for some specialized types of operations. Obviously immature ecosystem, expect a couple of years at minimum before adoption if it's the real deal.
354.
by VHRanger 3y ago
What's extraordinary is how ordinary massive trees used to be in North America. http://www.oldgrowth.ca/the-past-isnt-what-it-used-to-be/ Even in Ontario we used to have forests where a large portion of trees were
355.
by VHRanger 3y ago
Yes, there's a reason all the tallest trees recorded are a bit more than 100m high. That's basically the upper limit to get water to the leaves up top
356.
by VHRanger 3y ago
> That’s a tall order to avoid condensation. Oh, indoor condensation absolutely will happen. Even in a new build that's up to "passive house certification" levels of well-insulated. You just have to design around this trut
357.
by VHRanger 3y ago
Ah right it was decking 6x6 that warped in the sun. How do you prepare those after buying before building?
358.
by VHRanger 3y ago
Yeah, drying wood in the sun in one big reason for warping. Big box stores sell you wet wood - when you have it on the construction site it'll start warping and checking as it gets exposed to sun
359.
by VHRanger 3y ago
Because bamboo is really bad at many things we do with wood. Wood is different. Each type of wood has its advantage and disadvantage. A pine tree is more different to a Maple tree than you are to a great white shark in the evolution chart.
360.
by VHRanger 3y ago
I don't doubt Swiss made stuff is much tighter and has better detailing than North American builder grade crap. For the "leaks bugs" part - ladybugs here will find a way in as soon as you have a 1.5-2mm gap. They find them al

More ›