singularity2001

4 ms·

31.
by singularity2001 1mo ago
Also, why don't they bake in these limitations via reinforcement learning so they can keep the prompt context clear.
32.
by singularity2001 1mo ago
it's a paradox given how extremely complex each single neutron is. my current solution is that most of the complexity of neurons is just for maintenance
33.
by singularity2001 1mo ago
4000 BC? I don't buy it: if you look at how little Greek changed over the last 3500 years and compared to how different it is to other languages this doesn't sit right
34.
by singularity2001 2mo ago
in the US you may ask Siri about stuff in Europe it still takes about half a minute to activate GPT and talk to it. Investing $1000 in a new iPhone might solve the problem soon but at this point I'd rather try a new $300 device just ou
35.
by singularity2001 2mo ago
I don't want it to be a lab leak but all the evidence points towards it being a lab leak
36.
by singularity2001 2mo ago
there is a practical reason to get rid of the fantasy reals and restrict oneself to normal reals or some other new invention: Since Lean has become more popular as a proving system I've stumbled upon one very annoying feature of reals:
37.
by singularity2001 2mo ago
what he probably meant is that if you write down pi with 200 digits there's nowhere in the universe where you can find pi with that precision
38.
by singularity2001 2mo ago
World Models, really big world models like Dreamer V4, which unfortunately you can't do easily as a master student, but maybe you can work with tiny versions and prove that the direction works, or Lora some foundation model. If you loo
39.
by singularity2001 2mo ago
Have you tried writing any programs in it?
40.
by singularity2001 2mo ago
Maybe Terrence is chasing too many red herrings. Maybe he should just have asked it to find a counter example to the conjecture and on success explain how it was found.
41.
by singularity2001 2mo ago
You are badly informed the counter-example was found shortly before. Terrence just tried to replicate how it was found.
42.
by singularity2001 2mo ago
they're still bounded by training data and how it relates to the real world. Yes and no. They can extrapolate and build upon the training data, as was the case with the last dozens of math proofs
43.
by singularity2001 2mo ago
They will have decades of experience with AI and they will be able to guide them by sniffing their hallucinations from single words.
44.
by singularity2001 2mo ago
The Redox operating system is a complete rewrite of operating system concepts in Rust. Disclaimer, I worked on the Mac OS port.
45.
by singularity2001 2mo ago
> The other issue is Gen Z and Gen A are now very much opposed to AI. Jinxzi has Bernard as friend though
46.
by singularity2001 3mo ago
Exactly. It's a good idea, badly executed.
47.
by singularity2001 3mo ago
As the author mentions the Trojan War in the Illias could be reflections of this event
48.
by singularity2001 3mo ago
Very good question I can only answer for one subset tracked by Terence Tao https://github.com/teorth/erdosproblems
49.
by singularity2001 3mo ago
If anything their capability of abstract inductive reasoning is way beyond the average human given how much better LLMs are at solving math problems, it's the paradox that they can do complicated reasoning before they can do intuitive
50.
by singularity2001 3mo ago
Agents have a state which will unfold as a plan, especially in planning mode. Why not call this 'will'?
51.
by singularity2001 3mo ago
Now "Jetlag the game" HAS to shoot their next episode over there
52.
by singularity2001 3mo ago
Words are also not necessary for LLMs, hear me out: they could be trained mostly on video showing how to solve problems and would work almost the same (in principle). That's because both word tokens and video will end up in the same la
53.
by singularity2001 3mo ago
Because these models are now also trained on visual data, so they have a common abstract language in the latent space for different kinds of modalities. It's perfectly reasonable to ask the model if it can associate an image of a serve
54.
by singularity2001 3mo ago
I thought in this day and age "world model" also includes robo arm training data and robot arm benchmarks
55.
by singularity2001 3mo ago
Wait, I've seen the same libation formula appearing in the Phaistos disc. For those 10 of you who have the fonts installed: 𐇑 𐇘 𐇪 𐇐 | 𐇬 𐇳 𐇖 𐇗𐇽 | 𐇬 𐇗 𐇜 | 𐇬 𐇼 𐇖𐇽 | 𐇥 𐇬 𐇳 𐇖 𐇗𐇽 | 𐇪 𐇱 𐇦 𐇨 | 𐇖 𐇡 𐇲 |…
56.
by singularity2001 3mo ago
If it turns out to be true, it would open the door a bit for connecting Indo-European languages with Semitic languages. In the beginning of the last century it was believed that these were related. Later this came out of vogue. How could th
57.
by singularity2001 4mo ago
Most importantly, the reinforcement loop is used during training. I don't agree with Sutton's original hypothesis, but it holds even less after reinforcement learning.
58.
by singularity2001 4mo ago
or could it be the other way around that actual privacy is forbidden in Europe because they want to read your messages
59.
by singularity2001 4mo ago
do they not have data centers in Europe yet
60.
by singularity2001 4mo ago
some of the greatest ideas are proposed in a ridiculed manner first

More ›