bubaumba
bubaumba on Hacker News
> L40S are great cards and very cost effective
from https://www.asacomputers.com/nvidia-l40s-48gb-graphics-card....
nvidia l40s 48gb graphics card
Our price: $7,569.10*
Not arguing against 'great', but cost efficiency is questionable. for 10% you can get two used 3090. The good thing about LLMs is they are sequential and should be easily parallelized. Model can be split in several sub-models, by the number of GPUs. Then 2,3,4.. GPUs should improve performance proportionally on big batches, and make it possible to run bigger model on low end hardware.
it may be even cheaper to use big models through API. Claude, GPT, whatever. Rental is efficient only for big butches, while API is priced per call/size and is cheap for small models.
It's a trick. If you donate something to charity you can write it off the taxes. After IPOs most new billionaires did it, including Zuk. It's enough to just promise it publicly. There is no time limit, it can take years to 'donate', but writing off is immediate. To optimize it even further they create their own dummy charities full of friends and relatives.
Simple tax optimization. Like new billionaires promising to significant donations. a) they don't have to donate. b) they can immediately slash those promised donations from taxes.
Absolutely worth doing. Here is interesting related video, local RAG:
https://www.youtube.com/watch?v=bq1Plo2RhYI
I'm not an expert, but I'll do it for learning. Then open source if it works. As far as I understand this approach requires a vector database and LLM which doesn't have to be big. Technically it can be implemented as local web server. Should be easy to use, just type and get a sorted by relevance list.
3. Quick lookup into internal documents. Almost any company needs it. Navigating file-system like hierarchy is slow and limited. That was old way.
4. Quick lookup into the code to find relevant parts even when the wording in comments is different.
This is cool, but how about local semantic search through tens of thousands articles and books. Sure I'm not the first, there should be some tools already.
> I don't think I see them as a win
Unavoidable, probably
> but they're easily dealt with. AI will need analysts at the latter stage to evaluate the outputs but that will be a relatively short-lived problem
That solves only to some degree. Hallucinations may happen at this stage too. Then either correct answer can get rejected or false pass through.
> it will just have much better precision than us.
and much faster with the right hardware. And that's enough if AI can do in seconds what humans takes years. With o3 the price is only the limit, looks like.
After collision drone should shut itself down and drop like a rock. Looks like it wasn't the case here. And there should be nobody on the ground below, of course. Flying over that lake would be safe
> home made explosive
With this a lot of damage can be done even without drone. As for weaponized it's not a future, it's a reality in Ukraine for years now. Defense against them is difficult to impossible. A bodyguard who can sacrifice himself may sometimes work.
Nothing to worry about. They are dead end only if we find something better. Till then they here to stay. And likely even after they will be used.
They are not mutually exclusive. Likely we'll get more clear separation of architecture and underlying technology. In this case agents (i.e. architecture) can use different technologies or mix of them. Including 'AI' and algorithms. The trick is to make them work together.
> AGI is special. Because one day AI can start improving itself autonomously
AGI can be sub-human, right? That's probably how it will start. The question will be is it already AGI or not yet, i.e. where to set the boundary. So, at first that will be humans improving AGI, but then... I'm afraid it can get so much better that humans will be literally like macaques in comparison.
> if the underlying LLMs dont get any better, there is no reason to expect the system built out of them to get any better.
Actually o1, o3 are doing exactly this, and very well. I.e. explicitly: by proper orchestration the same LLM can do much better job. There is a price, but...
> you would expect to be able to build agentic systems out of much smaller LLMs
Good point, it should be possible to do it on a high-end pc or even embedded.
> search speed for potential room temperature superconductors?
and what if it's a dead end?
> My brother and I both read around 900 wpm.
That's a great speed. I'll never reach because I have to read some paragraphs several times before I understand the article. Some times using pen a paper...
Anyway, you can still improve you speed with the same level of understanding by asking chatgpt for summary.
> multiple years in with much investment and little return
Hmm.. where do you live? It's quite transforming already in many areas and not going to stop any time soon. I'll call it 'The Last Explosion', as opposite to 'AI Winter'. By that I mean this explosion with result in AGI. Likely sub-human first, then super-human.
Like it or not Meta is a major player in AI world with its free models and tools.
As for social impact of the rest it's debatable. I personally don't have active social accounts, and not sure this is good.
I can explain a bit. Russians living in Empire of Evil can see all internet including US and EU news. At the same time 'Putin propaganda' channels are blocked in EU. In EU only one side is available. This creates an information bubble, as intended. Which is a basic crowd control technique used to drive public opinion. In this case to support the war. The result is obvious, EU polls show much stronger support than the rest of the world. Even though media claims most of the world is against Putin, if you look at the map it's only minority, NATO and a few allies. In some EU countries it's even a crime look through the bubble's wall. Most don't realize it even exists. They accept the arguments from their politicians. Like it's a business opportunity, or it's a cheap way to harm Putin. The price for that is hundreds of thousands of human lives on both sides. Which is generally considered as ok, as those are Russians and Ukrainians, not us. Actually media doesn't talk much about it.
> European press agencies
It's very naive to believe in 'European press'. To get the idea check Ukrainian war coverage. What you'll see first is how single sided it is. This cannot be a coincidence. It can be only a result of total control. I respected 'The Guardians' before, but after eyes opening it appears to be the most brainwashing and manipulative there. Very professionally done, must admit. The problem isn't just that war, it's likely everything and I have no easy way to check for example what really happened in Afghan war. Did US really won like Biden said?
You may think Anthropic is different? It's the same UI + API. The difference is they don't create free accounts for personal emails. Sort of shadow banning. However they accepted my work email for free account and personal email for payed API. May be that's just me, I tried several personals.
As for model's quality they are both impressive. I haven't run meaningful comparison tests.
They are both. and there are UIs for their API, which costs $$. But web UI is free. And limited. It's nice to have options.
it's impossible to do it safely. open source model for kids!? without moderation every 'hacker' would like to get access. imho, the only option is to sell support as a business to someone else. with payed subscription it could survive for some time. but obviously not for long without new robots ans subscriptions. more interesting would be to have this sort of chatgpt for kids. with some basic robot control like flashing lights. then it would be possible to have several small and big manufacturers making mechanical robots and contributing to common software development.
another proof for 'robotics is hard'...
it should be possible to ask model to think aloud (or step-by-step) and then give summary. in one or two prompts. give only summary back to user.
> and aren't nearly as rude as modern Americans
You haven't been in Russia or post-USSR.
May be not, it could be just new normal. Reading old books I noticed how nicely people are talking to each other. But it may only sound nice by modern standards while being actually the same by old.
BTW, it's easy to test model's logic and truthfulness by giving it a wrong decision is if it was its, and asking to explain. Model has no memory and cannot distinguish the source of the text. 'Truthful' model should admit mistake without being asked. Likely model instead will do 'parallel construction' to support 'its' decision.
> Also, it's advertisers that stay away,
They are coming back and line up