SpaceManNabs

18 ms·

91.
by SpaceManNabs 1y ago
I am not surprised. Every time I mention the draconian laws around digital speech when flying into london, hackernews historically said I was being ridiculous. The UK has some of the oddest laws I have seen from a western nation.
92.
by SpaceManNabs 1y ago
Thank you for spending your time to answer my questions.
93.
by SpaceManNabs 1y ago
I thought one of the draws for apple devices was privacy. Wonder how they account for that. In addition, i know a lot of apple device owners dislike AI integration. For example, my non tech friends turned off a lot of the apple intelligence
94.
by SpaceManNabs 1y ago
thanks! got any links if you can spare the time? i think the info on gans using transformers might be enough. wasn't aware!
95.
by SpaceManNabs 1y ago
I miss GANs. I understand that they are much harder to train than transformers for the same performance even with high data regime and high parameter regime, but there was such good optimization research and tricks that came out of them. Th
96.
by SpaceManNabs 1y ago
not affiliated with them and i might be a little out of date but here are my guesses 1. prompt caching 2. some RAG to save resources 3. of course lots model optimizations and CUDA optimizations 4. lots of throttling 5. offloading parts of t
97.
by SpaceManNabs 1y ago
So are foundational models real finally now? Are they just multimodal for everything? Are foundational time series models included in this category?
98.
by SpaceManNabs 1y ago
> Growing up half in England and US I feel British culture is more attuned to the class aspects to this kind of event. Traditionally America likes to pretend this kind of class contempt doesn't exist (think of, people on welfare ang
99.
by SpaceManNabs 1y ago
If Japan wants to limit tourists, they can. If they don't, it is because they truly don't want the tradeoff.
100.
by SpaceManNabs 1y ago
Isnt there masked diffusion as well?
101.
by SpaceManNabs 1y ago
one of my interns didn't know how to create a folder.
102.
by SpaceManNabs 1y ago
That isnt my claim. My claim is that they have access to different reaources. Whether that makes a difference it seems not but maybe it does.
103.
by SpaceManNabs 1y ago
Why is it not a fair assessment to say it is comparing two "clients" with different resources if one can do a web lookup and the other cannot? You note yourself that it was meaningful in another round. > Also, the web search wa
104.
by SpaceManNabs 1y ago
> Looking forward to the inevitable goalpost-moving of "that's not real reasoning" It did a web lookup. It is not comparing humans and o3 with equal resources.
105.
by SpaceManNabs 1y ago
The article does a wonderful job in providing context for the proof. I really enjoyed the clear descriptions of the three scales.
106.
by SpaceManNabs 2y ago
> Also — secondhand books are generally outdated, undesirable, and/or damaged — do collectors still find diamonds in the rough? Growing up poor, second hand book stores were how I found some of the best math books that I wouldn&#x27
107.
by SpaceManNabs 2y ago
> someone found the good interesting enough to buy it in the first place (1st curation) I wonder how this is offset by someone finding an item not worth enough to keep. But maybe I underestimate the generosity (or need for income) of the
108.
by SpaceManNabs 2y ago
I have found that second hand shops tend to have a more diverse and rich selection and charity shops can have very rare gems. Support both, and I wouldn't want to give up one for the other. Also a bit of an aside, but charity shops are
109.
by SpaceManNabs 2y ago
i dont think those demos are comparable and cool for sharing your link!
110.
by SpaceManNabs 2y ago
Which documentary? Is it this one? https://www.youtube.com/watch?v=UuhECwm31dM
111.
by SpaceManNabs 2y ago
I applaud you for acknowledging your mistake. So many people double down, especially in this pernicious and polarized age.
112.
by SpaceManNabs 2y ago
> Yeah it's surprising that it works from such sparse rewards. I think imagining a lot of scenarios in parallel using the world model does some of the heavy lifting here. This is such gold. Thanks for sharing. Immediately added to m
113.
by SpaceManNabs 2y ago
Let's look at 2025 https://www.youtube.com/watch?v=x-exzZ-CIUw It looks pretty awesome. Let's see what happens.
114.
by SpaceManNabs 2y ago
> And that is why it is never going to work in the real world: games have clear objectives with obvious rewards. The real world, not so much. I encourage you to read deepmind's work with robots.
115.
by SpaceManNabs 2y ago
I just want to express my condolences in how difficult it must be to correct basic misunderstandings that can be immediately corrected from reading the fourth paragraph under the section "Diamonds are forever" Thanks for your hard
116.
by SpaceManNabs 2y ago
If you have an alternative for RL in these use cases, please feel free to share. When RL works, it really works. The only alternative I have seen is deep networks with MCTS, and they are quickly to ramp up to decent quality. But they hit ca
117.
by SpaceManNabs 2y ago
This thread is all you need to know that PG types are ok with losing 10-50% of their equity's value as long as they can play tribal politic games. PG is on X every day pretty much going "Why is trump doing this? There must be a re
118.
by SpaceManNabs 2y ago
> what strikes me as particularly insane is how it’s not even defensible as a protective measure You must not have read many of the comments here. Way too many people are trying to defend this just because they don't want to have to
119.
by SpaceManNabs 2y ago
This is a bunch of nonsense that is not driven by data and not even worth a response. I am tired of people spitting BS just to defend Trump's inane policies and everyone else having to rederive modern economic trade theory in comments
120.
by SpaceManNabs 2y ago
This is blatantly false. You just have to look at Jerome Powell's reasoning in 2018-2019 and just last month!

More ›