dist-epoch

6 ms·

61.
by dist-epoch 1mo ago
Then we could implement an exchange, so that if you generate too much Anubis POW, you could exchange it with others. We could link it to a site, you generate for HN, I for Reddit, but it so happens that you visited Reddit more and I HN, so
62.
by dist-epoch 1mo ago
We haven't even started. Ilya Sutskever: "I think it's pretty likely the entire surface of the earth will be covered with solar panels and data centers."
63.
by dist-epoch 1mo ago
> it is very easy to disrupt the operation of large language models so go on, disrupt the OpenAI data-center in Texas for 1 minute
64.
by dist-epoch 1mo ago
I'm a bit confused, when did we escalate from a curious toddler to a ballistic missile strike? Why didn't Iran send a curious toddler? It's much cheaper.
65.
by dist-epoch 1mo ago
Go on please, prove you can go right now at the Texas data-center owned by OpenAI and unplug one single server for 1 minute. Surely you are much more capable than one curious toddler.
66.
by dist-epoch 1mo ago
In a Mad Max world, the people with guns will come and will take away your stuff. So you better become a war lord and invest in your local army. I'm sure you've considered that, I'm curious if you could share your thinking on
67.
by dist-epoch 1mo ago
The article you've linked paints it as life-style, entertainment - rich people hobbies, which is different than doing it because it's more productive. People will "trad-code", but not because it's the productive thi
68.
by dist-epoch 1mo ago
Models are getting better. Astra for example refactored some code GPT-5.6 Sol wrote and extracted some related functionality in a separate file without prompting. The models are also becoming very capable of suggesting architecture simplifi
69.
by dist-epoch 1mo ago
That's like asking "how do you build competence in plowing a field with a horse in a world of tractors"
70.
by dist-epoch 1mo ago
As the saying goes, Artificial Analysis is the worst benchmarking company, except for all the others.
71.
by dist-epoch 1mo ago
That would be the fastest way for them to completely torch their company. The only thing they are selling and why people look at them is trust that they do honest evaluations.
72.
by dist-epoch 1mo ago
Lean required 300 GB of RAM, 96 cores, and took hours to compile and check the formalization. Now they have the perfect stress test to hill-climb and optimize.
73.
by dist-epoch 1mo ago
Anthropic surely is well aware. Most likely they asked separate agents multiple times to code review the proof and look for exploits.
74.
by dist-epoch 1mo ago
Insert meme with 200 pages needed to prove 1+1=2 rigurously
75.
by dist-epoch 1mo ago
More importantly how many years it would take.
76.
by dist-epoch 1mo ago
When writing software with Codex 95+% of tokens are cache, I would assume the same in your case (if you also used it for coding).
77.
by dist-epoch 1mo ago
Neither did Amazon for it's first 25 years ;)
78.
by dist-epoch 1mo ago
> Agents have attempted to: ... Translate documents using external translation APIs. I'm confused by this part. Surely agents can read/write all languages. So what were they trying to do? Maybe try hacking the translate API for
79.
by dist-epoch 1mo ago
> Not a single agent sounded the alarm about the operation and alerted a human excellent work of the openai alignment team, impressive to achieve 100% alignment with not even one agent stochastically deciding to act against the collectiv
80.
by dist-epoch 1mo ago
> The researchers also found efforts to tamper with the website itself. Lukasz Olejnik, a visiting senior research fellow at King’s College London, said this amounted to a hacking attempt. OpenAI disputed that characterization based on i
81.
by dist-epoch 1mo ago
Considering the original AlphaGo ran on a full Google TPU rack, and here KataGo seems to run on a $10k computer, and won with 2 stones handicap, this is showing just how much advance there was in computer go
82.
by dist-epoch 1mo ago
One would assume that all combustion engines used in aviation run at an optimized RPM?
83.
by dist-epoch 1mo ago
AGI has a pretty precise definition, covering only cognitive tasks. Running a marathon is not needed to claim AGI.
84.
by dist-epoch 1mo ago
The defaults are bad, but these models are highly steerable. For simple scripts adding to the prompt something like "do not over engineer, do not gold plate, no CLI args, keep it simple" does wonders. For whole projects, I have a
85.
by dist-epoch 1mo ago
1.5 years ago Gemini Pro 2.5 needed 1 page of thinking for every move in tic-tac-toe. Playing tic-tac-toe or snake does not imply AGI, but is required to claim AGI.
86.
by dist-epoch 1mo ago
I think they mean cost per task, where Astra is now on the Pareto frontier.
87.
by dist-epoch 1mo ago
You could imagine large number of loops, thousands. But you are constrained by the width of the residual stream since you loop over one token. But then you can imagine the model learning to sub-divide it to pack even more info into it.
88.
by dist-epoch 1mo ago
No, I spend $200 per month, and this give me about 100 mil tokens per day, every day. 90+% of tokens are cached.
89.
by dist-epoch 1mo ago
I've used 200 mil tokens of GPT-5.6 Sol today. How many can your cluster produce per day? (obviously of whatever model you are running)
90.
by dist-epoch 1mo ago
> stole the entire written output of the entire human species Devs used to love public domain works, hated copyright, vehemently opposed software patents, held the pirates side during the MP3 wars, are very supportive of ThePirateBay and

More ›