p1esk

19 ms·

421.
by p1esk 2y ago
I'm talking about web interface, not API. Should be available now, since they said "immediate release".
422.
by p1esk 2y ago
More like you will need to dip into your 401k fund early to pay for it after they raise the prices.
423.
by p1esk 2y ago
Do people see the new models in the web interface? Mine still shows the old models (I'm a paid subscriber).
424.
by p1esk 2y ago
after weighing multiple factors including user experience, competitive advantage, and the option to pursue the chain of thought monitoring, we have decided not to show the raw chains of thought to users.
425.
by p1esk 2y ago
They also show ads: https://www.researchgate.net/marketing-solutions As a researcher I don’t see any value there. I’ll stick with Arxiv, thanks.
426.
by p1esk 2y ago
It would be a lot more convincing if you actually ran it yourself and did a proper apples to apples comparison, especially considering that’s the whole idea behind your project.
427.
by p1esk 2y ago
Then how did you compute the 30-70% cost efficiency numbers compared to Nvidia if you haven’t run this Llama finetuning task on Nvidia GPUs?
428.
by p1esk 2y ago
Loan officers typically want to see bank account balance, paystubs, and tax records.
429.
by p1esk 2y ago
Did you try running this task (finetuning Llama) on Nvidia GPUs? If yes, can you provide details (which cloud instance and time)? I’m curious about your reported 30-70% speedup.
430.
by p1esk 2y ago
people's brains are already pretty good at reconstructing 3D from 2D But we do prefer stereo over mono in audio domain, right?
431.
by p1esk 2y ago
What does she do with it?
432.
by p1esk 2y ago
I see, OK, I misinterpreted it as per node bandwidth. Yes, this makes more sense, and is probably fast enough for most workloads.
433.
by p1esk 2y ago
I thought NVLink is only for communication between GPUs within a single node, no? I don't know what the size of their switches are, but I verified that within a 64 node cluster I got the full advertised 3.2Tbps bandwidth. So that'
434.
by p1esk 2y ago
LambdaLabs GPU cluster provides internode bandwidth of 3.2Tbps: I personally verified it in a cluster of 64 nodes (8xH100 servers) and they claim it holds for up to 5k GPU cluster. What is the internode bandwidth of Frontier? Someone claime
435.
by p1esk 2y ago
200 Gbps Oh wow, that’s pretty bad.
436.
by p1esk 2y ago
Do you mean this supercomputer has slower internode links? What are its links? For example, xAI just brought up 100k GPU cluster, most likely with 800Gbps internode links, or maybe even double that. I think the main difference is in the tar
437.
by p1esk 2y ago
Am I misreading something? Yes, you are. NVM stands for 'non-volatile memory", which is literally the opposite of DRAM. Analog computation can be done using any memory cell technology (transistor, capacitor, memristor, etc), but
438.
by p1esk 2y ago
You don't need to write anything back to flash to use it to compute something: the output of a floating gate transistor is written to some digital buffer nearby (usually SRAM). Yes, it's only used for inference, not sure how that
439.
by p1esk 2y ago
Let’s see: Mythic uses flash, d-Matrix uses SRAM. Encharge is the only one who uses capacitor based crossbars, but those are custom built from scratch and very different from any existing DRAM technology. Which companies are using DRAM for
440.
by p1esk 2y ago
In-memory doesn’t mean in-DRAM. https://arxiv.org/pdf/2406.08413
441.
by p1esk 2y ago
They're just a self-assured and competent trades-person that's easy to work with and anticipate value from. Such people are rare. I’m speaking from experience working at multiple startups. It’s even worse at big companies.
442.
by p1esk 2y ago
Someone will surely set its objective for survival and evolution.
443.
by p1esk 2y ago
After many years in CA I love it in Florida. Palm Beach or Boca Raton are very nice, schools are excellent, weather is perfect 7 months out of 12. And houses are significantly cheaper than in equivalent areas in CA. Less woke culture too.
444.
by p1esk 2y ago
[flagged]
445.
by p1esk 2y ago
sunlight takes about 30,000 years to work its way out from the centre to the surface of the Sun What? Why?
446.
by p1esk 2y ago
[flagged]
447.
by p1esk 2y ago
Average work hours for CEO (according to the article): 62.5/week. I’m guessing that’s also true of those who are trying to become CEO. Personally, I’d rather not spend almost all my waking hours at work, most of my life.
448.
by p1esk 2y ago
And it means something else in Hollywood. But we are discussing language models here, aren’t we?
449.
by p1esk 2y ago
No, I mean, a transformer is a very specific model architecture, and your simple language model has nothing to do with that architecture. Unless I’m missing something.
450.
by p1esk 2y ago
Why do you call your language model “transformer”?

More ›