zozbot234

8 ms·

121.
by zozbot234 4mo ago
1HP sustained is not insane for a horse. A human OTOH is a very different matter.
122.
by zozbot234 4mo ago
Buying one's way in doesn't exactly negate being interested in the underlying venture. He still had to provide significant funding.
123.
by zozbot234 4mo ago
> The intentions of your garden-variety nerd talking about atheism are pretty clear, and it's not to make some greater theological point. I agree about the underlying intentions, but I was talking about the typical, literal argument
124.
by zozbot234 4mo ago
> Try running the latest OS models on a normal Mac or PC. It can be done through the magic of SSD offload. The worst case involves seconds-per-token speeds, but that's OK if you only care about low volumes of slow unattended infere
125.
by zozbot234 4mo ago
Not for proprietary models, all you get is a terse summary.
126.
by zozbot234 4mo ago
Among other things, because you simply can't get those "massive amounts" of text from a SOTA model at reasonable cost. And complex reasoning cannot possibly be trained in a pure one-shot fashion, real post-training takes mass
127.
by zozbot234 4mo ago
"Distillation" from APIs is not a thing, it cannot replicate a model's deep reasoning and behavior.
128.
by zozbot234 4mo ago
If "everyone understands religion is not literal", why do so many people take it literally? You could just as sensibly flip the argument and argue that the garden-variety 'nerdy' atheist is talking literally about atheis
129.
by zozbot234 4mo ago
That's a lot of money out in rural country, which is where veterinarians tend to be especially useful.
130.
by zozbot234 4mo ago
I think that plenty of 'nerds' would very much disagree about that. Steering agents effectively is something that can take massive amounts of creativity and talent and be quite helpful wrt. the final result. There's no real a
131.
by zozbot234 4mo ago
That's an interesting point since in practice, athletic ("jock") women tend to be more common than nerdy women to begin with. Women being into tech or science (stereotypical "nerdy" pursuits) is something that tend
132.
by zozbot234 4mo ago
There's plenty of wealthy folks who aren't especially evil or greedy in any real way, but you wouldn't know that because they don't tend to show off or spend their wealth to begin with, they just shepherd it and grow it
133.
by zozbot234 4mo ago
> Musk for me was never a nerd. Many "founders" aren't nerds for me. It's pretty hard to describe Elon Musk's ventures in space exploration, robots and human-like AI as anything other than prototypical, "cor
134.
by zozbot234 4mo ago
It's weird to describe this as "employment" when many and perhaps most drivers these days are simultaneously "employed" by both Uber and Lyft, serving rides for both. This kind of informal relationship is a lot clo
135.
by zozbot234 4mo ago
> Let's put this math in a mirror and do the leftie version of the exponentials The plot twist is that the 'rightie' and the 'leftie' are both entirely correct. Which is why most developed economies try to remov
136.
by zozbot234 4mo ago
One in 500 of them makes themselves a billionaire, the rest have thrown two million dollars down the toilet. It's just a fair bet, there's nothing "rigged" about it.
137.
by zozbot234 4mo ago
If the only viable alternative is that nobody gets to start their business rather than some idealized best case of everyone getting a chance of it, then yes, it's probably fair.
138.
by zozbot234 4mo ago
There's no resource destruction involved in displacing taxi drivers. Taxi medallions are a rent-seeking scheme, there's no real scarcity involved. Most other instances of "creative destruction" are like that, the capit
139.
by zozbot234 4mo ago
For sparse MoE models, the single expert layers that the inference gets sampled from are actually quite small - single-digit megabytes or so.
140.
by zozbot234 4mo ago
Normally, experts are picked for every layer not just every token. But there are plausible ways of getting around that bottleneck while streaming if you can batch many inferences together. Still, the Apple approach of swapping the experts
141.
by zozbot234 4mo ago
I don't think educational "proof of concept" code can be described as even loosely realistic cyber offense in this day and age. The Mythos preview paper claimed an ability to stage attacks in an end-to-end fashion and work ar
142.
by zozbot234 4mo ago
> First of all I found that fable is trained in a way that even if you were to jailbreak it, it would be completely uninterested in exploitation or finding creative solutions for explotation. This is quite relevant if true. People have
143.
by zozbot234 4mo ago
DeepSeek V4 Pro is ~800GB total at native quantization (1.6T params with most being 4-bit) so it can run on the hardware you mentioned. There is also a 2-bit version that will run on a single 512GB machine. SSD streaming also makes lower-e
144.
by zozbot234 4mo ago
You need a lot less than that if you're willing to stream the model from SSD. At that point, the best machine is probably a cheap old-gen HEDT with lots of PCIe lanes to attach cheap NVMe storage to, so as to stream the model at reason
145.
by zozbot234 4mo ago
You can use batching on consumer hardware, it just requires a KV-cache efficient model (or short context only) and keeping multiple inference flows running in parallel. This is most useful in combination with streamed inference, since the
146.
by zozbot234 4mo ago
You forgot to mention that it takes a lot more energy to train that human before they're able to work.
147.
by zozbot234 4mo ago
Knowledge-base access is not very useful in general because a model doesn't have well-defined "known unknowns" that might trigger an agentic search of the outside knowledge base. Plus surfacing knowledge you don't know m
148.
by zozbot234 4mo ago
If that's Kimi-based it would very much be on the larger side of open-weight models (1T params).
149.
by zozbot234 4mo ago
DGX Spark is effectively prosumer hardware, better than most consumer stuff but still not comparable to actual datacenter gear. You can't just look at TDP in isolation without also comparing performance.
150.
by zozbot234 4mo ago
Note that warning shot in that blog post means specifically a near-disaster event (perhaps one that's just barely averted) that's specifically caused by the AI. So far we've had AIUI no significant indication of open-weigh

More ›