tjwebbnorfolk

3 ms·

31.
by tjwebbnorfolk 2mo ago
It's already been trained. GDM said it was going to be released and it hasn't yet. Not sure why so much snark
32.
by tjwebbnorfolk 2mo ago
where's gemma 124b-a15?
33.
by tjwebbnorfolk 2mo ago
Nevertheless, experiments are necessary. Political entrepreneurialism is necessary if you want anything to improve. You can experiment on everyone, or you can experiment at the local level. One is obviously preferable over the other. Nobody
34.
by tjwebbnorfolk 2mo ago
I mean, there's not like some "big silicon" conspiracy keeping people from homelabbing their own chips. I suppose we'll find out at some point, but it sure as hell isn't possible in 2026 AD.
35.
by tjwebbnorfolk 2mo ago
More meta derangement syndrome on HN, what a surprise. We all benefit when companies invest their resources in producing open models. No one thinks this absolves anyone of being terrible elsewhere. But we can still be happy about it.
36.
by tjwebbnorfolk 2mo ago
Well, my ancestors would already say that our scarcity is effectively zero. Yet we still have political arguments about healthcare and I get irritated when my AI service says I'm out of quota, two things that didn't exist 4 or 5 g
37.
by tjwebbnorfolk 2mo ago
yea right now local models are like the swing producer. When claude throws 503s or 429s at me, or I run out of quota, I fire up a local model. It's also resilience. I can still do AI stuff if the internet goes down.
38.
by tjwebbnorfolk 2mo ago
> must attack things like the very concept of trillion dollar semiconductor fabs Computers used to be the size of these fabs. Now we have them in our pockets. Not because we need to attack or rebel or something. It's because a lot o
39.
by tjwebbnorfolk 2mo ago
> Economics itself is the enemy. Economics arise out of scarcity. Scarcity is a timeless fact of life in a finite world, it's not some abstract invention by "economics". The problem isn't that more people want more th
40.
by tjwebbnorfolk 2mo ago
I put together a nationwide (US) land parcel dataset over the past year and a half, and have been building a bunch of apps on top of this data: https://landstats.com
41.
by tjwebbnorfolk 2mo ago
Yes it's well known that luxury goods do not fit this model. GPUs are capital goods. It isn't hypothetical when you understand the difference in the demand curves for different types of goods.
42.
by tjwebbnorfolk 2mo ago
Rationing works when there are alternative goods available. Spam instead of steak, in your example. One of the most difficult shortages during WW2 was rubber, because there was no (known, at the time) comparable alternative. With GPUs, rat
43.
by tjwebbnorfolk 2mo ago
When demand exceeds supply you have two options: make the people who value those goods the highest compete with each other to pay the natural market price for them, or have a lottery with price ceilings. Price ceilings create inefficiency,
44.
by tjwebbnorfolk 2mo ago
They are, because every university (in the US) takes anywhere from 10-50% off the top of every grant
45.
by tjwebbnorfolk 2mo ago
Inefficient per watt, yes, but local inference capacity is greatly underutilized in aggregate. If a model can run on a machine that already exists, that's a bunch of additional chips that don't need to be built.
46.
by tjwebbnorfolk 2mo ago
How many lavishly-paid deans and bureaucrats and administrators are employed for each grad student?
47.
by tjwebbnorfolk 2mo ago
do you have a source for their GB300 count?
48.
by tjwebbnorfolk 2mo ago
The point is you need to TRY in order to have a chance of finding validity. No one is talking about murder here, which a pretty trite counterexample. In the case of Apple and OpenAI, since that's actually what we're talking about:
49.
by tjwebbnorfolk 2mo ago
Looking at things from multiple perspectives and trying to see validity in each point of view is a virtue, not a vice. I really hate that "both-sides" has become a copy-paste insult by ideologues.
50.
by tjwebbnorfolk 3mo ago
> Companies that fail to comply can face fines of $200 per day for each affected Californian. Does this cover things like credit reports/scores? If someone submits a request to this DROP thing, is it possible data gets deleted that
51.
by tjwebbnorfolk 3mo ago
> ~60–76 s/token I don't know if I'd call this "running"
52.
by tjwebbnorfolk 3mo ago
> where sufficiently capable models may be able to quickly weaponize pandemic-level viruses with widely available materials if someone figures out a way to give an LLM full operational control over a virus lab, we've got a whole dif
53.
by tjwebbnorfolk 3mo ago
I guess I need a disclaimer on every comment that says "climate change is obviously real you can save your downvotes" jfc
54.
by tjwebbnorfolk 3mo ago
I guess I need a disclaimer on every comment that says "climate change is obviously real you can save your downvotes"
55.
by tjwebbnorfolk 3mo ago
what am I denying? some people were under the impression that wildfire is a totally novel thing in Spain. I'm only pointing out that fairly large wildfires happens every year, but that this one is larger than normal. I guess I need a d
56.
by tjwebbnorfolk 3mo ago
Yes that's what I meant by "this one is larger than normal"
57.
by tjwebbnorfolk 3mo ago
always has been
58.
by tjwebbnorfolk 3mo ago
What am I denying, exactly? This thread and the article pertain to El Nino. Is someone denying that El Nino exists?
59.
by tjwebbnorfolk 3mo ago
This thread and the article are discussing about El Nino, not global average temperatures.
60.
by tjwebbnorfolk 3mo ago
on average, 240k acres burn every year in Spain. this is somewhat larger, but not abnormal. it's just near people this time.

More ›