lossolo

35 ms·

571.
by lossolo 2y ago
O1 seems like a variant of RLRF https://arxiv.org/abs/2403.14238
572.
by lossolo 2y ago
It's because there is nothing novel here from an architectural point of view. Again, the secret sauce is only in the training data. O1 seems like a variant of RLRF https://arxiv.org/abs/2403.14238 Soon you will se
573.
by lossolo 2y ago
Cool, it looks similar to my crypto mining rigs (8xGPU per node) from around 7 years ago, but I used PCI-E risers and a dual power supply.
574.
by lossolo 2y ago
Not to mention the security/mitigation overhead of context switches.
575.
by lossolo 2y ago
Could you link your source of these trends? So I could see the numbers
576.
by lossolo 2y ago
I'm not sure it's a joke. A lot of data supports that, maybe not all of it is accurate but you see the trend. https://www.economist.com/science-and-technology/2024/06/12/...
577.
by lossolo 2y ago
One of Hedra job offers for ML role lists "diffusion transformers".
578.
by lossolo 2y ago
Which trends exactly? I have M1 PRO and I agree with OP.
579.
by lossolo 2y ago
China leads in 57 out of 64 technologies, up from just 3 twenty years ago. [1] The US lost its research advantage: it was leading in 60 technologies 20 years ago, down to just 7 today.[1] 1. https://www.aspi.org.au/report&#x
580.
by lossolo 2y ago
You can't criticize Israel or you will get punished by government in US. Where is your freedom of speech in this case? "The U.S. Supreme Court on Tuesday declined to revive a newspaper's challenge on free speech grounds to an
581.
by lossolo 2y ago
> I think he legitimately believes that “the woke mind virus” is an existential threat to our society Musk blames 'wokeness' for his daughter rejecting him after he didn't accept who she was. His entire conflict with the l
582.
by lossolo 2y ago
Denuvo has been disabled many times; however, the amount of work required to modify all the generated injection points is tedious—it's a LOT of work. It seems that fewer and fewer people are willing to spend weeks or months of their li
583.
by lossolo 2y ago
> it's just interacting with the author of the first benchmarksgame left me with a good impression of focus on fairness and level of detail you can draw conclusions how the basic aspects of a particular langue perform from IMO, usin
584.
by lossolo 2y ago
> It is more accurate to specify that Go favors the sweet spot of "smaller" amount of work ... focuses on maintaining consistent latency ... It seems like you wrote the same thing that I wrote, but you used different words for
585.
by lossolo 2y ago
> binary-trees benchmark is using Go 1.23.0 which is the latest version. Go submissions do not do anything differently compared to other GC-based languages for the most part. The fastest Java submission has interestingly much lower execu
586.
by lossolo 2y ago
If my experience differs from yours, where you claim that something illegal is happening, it obviously suggests that you might be using some kind of shady services. The only time I've had experiences like the one you’re describing is w
587.
by lossolo 2y ago
> This is a rather strong claim not supported by real-world data on how Go’s GC behaves under load The only thing you linked is an old binary trees benchmark from the benchmarks game, which is not equivalent to the workload in a typical
588.
by lossolo 2y ago
So, it’s anecdotal to your circle. Maybe you're just using some shady providers? Because that’s not my experience or the experience of those in my circle. You should be more careful when buying stuff online; this is where these things
589.
by lossolo 2y ago
> In Poland I don't even know the names, but if you ever leave your phone number at any doctor, dentist, or blood test lab, starting from next day you'll receive tons of phone calls offering "free products", invitatio
590.
by lossolo 2y ago
I see, thanks. Well, I've had similar problems, though not many. Maybe we just design things more carefully, but whenever we did encounter issues, knowing where the allocation happened was 90% of the investigation. The rest was just us
591.
by lossolo 2y ago
Is your opinion grounded in theory or practice? As a practitioner running Go in a high-volume, low-latency environment with tens of millions of allocations per second, we haven't found GC to be a problem. The median pause time is actua
592.
by lossolo 2y ago
Check this one https://www.unitree.com/g1/ this is SOTA.
593.
by lossolo 2y ago
This is a lot better link, with humanoid robot: https://www.unitree.com/g1/ From what I understand they started mass producing this and you can buy it from 16k USD. And here is h1 https://www.unitree.com&#x2
594.
by lossolo 2y ago
> The claims about memory fragmentation are less clear though As to memory fragmentation, memory allocations are grouped by size, so when objects of the same size are allocated and freed, the memory can be reused efficiently without caus
595.
by lossolo 2y ago
> This seems like the Kim Dotcom situation again. I'm not sure where this myth originated—perhaps from Kim Dotcom's Twitter account? I clearly remember the Megaupload case. They knew they were hosting pirated content, didn&#x27
596.
by lossolo 2y ago
Do you compete in some kind of a niche? There is OVH in EU.
597.
by lossolo 2y ago
> And lacks most of the advanced optimizations you'd expect from a modern GC. It can't move objects around, it doesn't have generations, it doesn't have good diagnostics (it can't even show you what are the GC ro
598.
by lossolo 2y ago
Sure, I did—a lot of them. These are the ones that were not in my training dataset in any form before I solved them, as it's impossible for a human to hold all scientific papers, historical facts, and in general, the entirety of human
599.
by lossolo 2y ago
> When an LLM solves a novel problem, it's also reasoning, unless you use some contrived definition of the word "reasoning" that doesn't match how the word is actually used in normal conversation. Also I fully expect
600.
by lossolo 2y ago
I was.

More ›