andy99

6 ms·

91.
by andy99 2mo ago
I had seen the cigarette. But but not the meat which costs $2,690 for four tuna-can size cans and is a mix of pork and chicken Thinking of a concept for a restaurant meal along the lines of a calibrated fried spam dinner, that costs ~2k for
92.
by andy99 2mo ago
Why not? If done right it’s a good way to talk about implications for those companies and provide some education like tailscale did here. We also saw Anthropic post about “our agent escaped too” and while I understand the incident caused th
93.
by andy99 2mo ago
Running local LLMs split across machines. Pretty sure the author has some other posts about doing that. Edit: https://www.jeffgeerling.com/blog/2025/15-tb-vram-on-mac-stu...
94.
by andy99 2mo ago
That’s just conflating reasoning and “parallel reconstruction” - there is such thing as reasoning, and I agree it’s probably less common in decision making, but it’s fundamental to many tasks where we figure out a solution, like writing an
95.
by andy99 2mo ago
Back in the day it was a bit of a cliche to bring up “clever Hans”, the horse that could do math, when talking about machine learning. He couldn’t do math but he read some cues from his handler of pick the write answers, the handler iirc wa
96.
by andy99 2mo ago
AI writing is not good. Seeing this post, and another comment on here earlier today to the effect of “we’ve solved prose”, I’m genuinely really confused. I don’t want to be insulting, even if the first place my mind runs is basically people
97.
by andy99 2mo ago
Seems they want the narrative to be that “Claude” (their computer program) independently attacked some organizations, ergo LLMs are dangerous etc. Another framing would be Athropic irresponsibly (vibe?) coded an attack script, and didn’t mo
98.
by andy99 2mo ago
Then maybe just this timing is really unfortunate, I think most people’s first reaction will be that it looks like a “us too” response to the OpenAI/hf thing.
99.
by andy99 2mo ago
Yes came here to say the same “look at us, our AI is also dangerous! Please ban our competition”
100.
by andy99 2mo ago
I would have said 0x to infinity-x. Some stuff I think it would be a good idea and waste a bunch of time and then give up and just write it myself. And like you say, there’s lots of stuff I would just never have done, basically any kind of
101.
by andy99 2mo ago
So, there is no subliminal learning in this situation, under what conditions would we expect it. I find a transfer attack to be a bit far fetched but it’s definitely interesting. If we trained from random initialisations on DeepSeek output
102.
by andy99 2mo ago
I didn’t see a gguf yet, going to be most interesting if there’s a quant that fits nicely into about 90FN so it can run in 128GB unified memory. It’s 12B active so should hopefully be pretty fast at 2 bit quant if it fits, as in the ratio o
103.
by andy99 2mo ago
There’s such an opportunity, it has never been easier to differentiate oneself now, all you have to do is ignore whatever online advice or LLMs say and do your own thing, it’s sad that almost nobody does it.
104.
by andy99 2mo ago
Depends what you mean by meal, the parent didn’t like pizza slices but I believe one of the thick granny slices at Outta Sight pizza is still under $10, I’d consider that a lunch. It’s a fair point that generally it’s focused on single item
105.
by andy99 2mo ago
Cheaper to just buy a 60 OZ Kirkland vodka and invert it into the existing water cooler come party time. Or these which sadly aren’t available anymore: https://www.cbc.ca/news/canada/edmonton/how-a-4-litre-jug
106.
by andy99 2mo ago
Looks like they updated the article to say 20,000 litres of additional fuel, it wasn’t like that before as what I posted above was a copy paste. Edit: it actually says at the bottom: (This story has been refiled to add the dropped wor
107.
by andy99 2mo ago
I’m glad he feels this way and is standing up for this, even if I didn’t agree with lots of metas practice in the past (they are the same company that wouldn’t let people have access to the original llama model - though to be fair Mark migh
108.
by andy99 2mo ago
> The first aircraft, which carries 20,000 litres of fuel and can seat 238 passengers Pretty sure a zero is missing right? Wikipedia says normal A350 capacity is ~150k litres if I’m looking at the right thing. 200k litres means almost
109.
by andy99 2mo ago
Why would it obviously give attackers the upper hand? If we accept the premise that AI helps find vulns more efficiently, shouldn’t this be at least equally helpful in securing against attacks and make that increasingly efficient?
110.
by andy99 2mo ago
Right, it’s really a foundation of post enlightenment society. These people, Dario et al, would have wanted to ban sharing information about calculus or Newtonian physics because of “safety” - it’s trying to go back to the dark ages where o
111.
by andy99 2mo ago
> If "Mythos-class" models are such a problem, then... why not just let it fix everyone's code? Because it doesn’t really confer the advantage they claim, especially compared to e.g. paying an equivalent amount of money to
112.
by andy99 2mo ago
Anyone who calls it “safety” probably has a certain world view and is more aligned with the big 2 (and stuck in 2023). There is a growing industry of commercially focused risk evals that has a broader customer base.
113.
by andy99 2mo ago
You forgot “what is the definition of ‘sufficiently capable’”. Presumably it’s anything that competes with Anthropic. If they’re around in a year, presumably they won’t care about Fable level and will only think that whatever competes with
114.
by andy99 2mo ago
> I want AI as an assistant that helps me make decisions, and ensures that I'm in the driver's seat. Right now we’re still in the benchmaxxing phase, I feel like most AI models of combination of models and harness are not reall
115.
by andy99 2mo ago
Definitely agree, at a big company anyway there is nothing to be gained by sending these, or by any communication not directly related to the process. It doesn’t come across as genuine, it’s neutral to slightly annoying and has no impact on
116.
by andy99 2mo ago
I think there’s a kind of laundering that goes on. Like if leadership just told devs to go build something (gave them a prompt) and the devs picked some defaults, leadership wouldn’t like it, they’d want some different interpretation of the
117.
by andy99 3mo ago
Nothing personal, if someone told me they’d generated a 15 page paper based on a two week LLM discussion I wouldn’t want to read it at all. My prior is that it would be the result of some sycophantic echo chamber that the LLM had reinforced
118.
by andy99 3mo ago
Just send me the prompt if you’re going to write with AI. No point in trying to convince me it’s great, the marginal cost for me to re-run the prompt is zero, just send the prompt - it’s like unzipping a file before you send it to me, it do
119.
by andy99 3mo ago
#1 in a very close race is way less useful when you have to walk on eggshells to avoid triggering censorship (“safeguards”) that either refuse or knock it down to another model. I’ve almost completely stopped using Claude (except some legac
120.
by andy99 3mo ago
You should assume the weights have been poisoned, the model has been prompt injected, etc and design software around it accordingly. You can’t really trust the model and it makes sense to always have controls around it and not depend on it

More ›