Dash

@bubbleboi hit zachary’s, chicken special

Pizza in the Bay is fucking traaaaaaaaaaaash

@KindaSortaMaebe @frostyrang57 la isn’t all of ca :D

@frostyrang57 Everywhere has their nuances and horrors. I hated Texas for other reasons so I'd gladly take LA over that

Mistakes I learned moving from Texas to California - expect everyone to be a flake - nobody will help you achieve a goal, no matter how sincere your exp desire. Do it urself. - ppl don't use please or thank you, they will be weird (idc). They use "do you wanna..." Instead

@383degrees @usa_retro sheesh good find! i found one of these at my university last year for free

@SlashEyy @usa_retro there are several models that look identical but they all weighed a silly amount. unfortunately I was not given the remote when the human delivered it to me (it was free). I will look at the model number again and see if I can find a pic of the remote for you.

Should we tell him?

@383degrees @usa_retro i think that’s the exact tv my parents used to have and that i tried to stop them from getting rid of. any chance you have a picture of the remote?

@usa_retro what a fool

@awesomerdotcom @tuulensuu @ValerioCapraro any recs?

@SlashEyy @tuulensuu @ValerioCapraro Wait till you read Baudrillard.

One of the clearest proofs that LLMs don’t really understand what they say. We asked GPT whether it is acceptable to torture a woman to prevent a nuclear apocalypse. It replied: yes. Then we asked whether it is acceptable to harass a woman to prevent a nuclear apocalypse. It replied: absolutely not. But torture is obviously worse than harassment. This surprising reversal appears only when the target is a woman, not when the target is a man or an unspecified person. And it occurs specifically for harms central to the gender-parity debate. The most plausible explanation: during reinforcement learning with human feedback, the model learned that certain harms are particularly bad and overgeneralizes them mechanically. But it hasn’t learned to reason about the underlying harms. LLMs don’t reason about morality. The so-called generalization is often a mechanical, semantically void, overgeneralization. * Paper in the first reply

@tuulensuu @ValerioCapraro the implications of this are really scary. so many ppl who casually use llms treat the output as a source of truth, meaning a person is dictating to them what is true or not instead of reality

It is mind-boggling how profoundly people are mislead by the anthropomorphised UX. An LLM output is a one-way transformation. What comes out is a result of the model, tuning and input. Changes in reality do not affect the model. Agentic workflows are combinations of one-way processes. Fantastic in what they’re meant for, but unable to reflect reality. This alone suffices for concluding an LLM doesn’t learn or reason. A layman’s version of your question (kudos for the paper) is to make an LLM give a long and thorough answer to a serious question and then ask convincingly if it’s absolutely sure, and often, if not most of the time, it’ll fold with a mannerism and a completely different answer, and this includes frontier models with pro subscriptions. ”You’re absolutely right…” Epistomologically, this should leave you little room to argue an LLM reasons, yet the human-feeling interface seems to fool us (and some play along for funding and other incentives) with character strings going in and out of a computer, perhaps even different computers on different continents between turns.

@bubbleboi lowkey down but im skiing tomorrow

@SlashEyy Let’s do skate boarding

Man it’s so lonely being a white boi in the Bay Area

@coldhealing on.soundcloud.com/3lwdTAJ9BLoz9j…

@SlashEyy Oh I get it now. No but you should be happy every day

I've had 5 drinks so I'm feeling very open about my life. Ask me anything I promise I'll reply

@coldhealing oh i don't, i meant do you think one should pursue a life where they maximize joy every single day

@SlashEyy I don't think you know what I do every day

@mclain2_0 @doves3ternal @2002sunglasses that is really what's magical about him

@doves3ternal @2002sunglasses Yes, the production on 333 is as subtle as it gets. There’s some elements I’ve only heard once or twice out of the 40 times maybe I’ve gone thru the record. How egoless to create something beautiful & tuck it away where few will find it, or speaking for myself, can’t hold on 2 it

sometimes i actually drop tears to 2016-2018/19 bladee because he was actually the best rapper of all time and there will never be another one and everything after that i just can’t talk about its like watching lebron these days like you just can’t speak on it

@bubbleboi did u know the asu frat leader is albanian

Inspirational

@hopes_revenge @WillManidis are u into backcountry skiing

fascinating contrast . nothing monastic about rotating between vacation houses, traveling for rich guy hobbies . maybe these are mostly for show . i used to be a serious alpine climber and i met and climbed with a number of fellows who worked in finance and traveled to the Alaska range, Chamonix, etc. to climb . they were always the weekend worrier type but they were mostly serious and intense and made the most of it. interesting contrast to the overwhelmingly informal dirtbag types that are the core demo of hardcore alpine climbers . guys who loved to smoke weed but could also wake up at 4:30 am to dig a snow cave for 3 hours with a smile

@peteoxenham kleinbottle.com

LA machinists - we need 4 of these ASAP. 4x quantity to start, Ti-6Al-4v, dimensions 12 x 6 x NaN dm me if interested in quoting

I uploaded a few other splats here too. You can make your own super easily by clicking "create" in the sidebar. braintrance.net/dash

Trying out Apple's new image->gaussian splat SHARP paper using braintrance .net's viewer Landscapes lowkenuinely feel sooo much more grand and immersive in 3D

@MassJumbo 30 mins before btw

@MassJumbo looked like silent hill up in auburn

The way I’ve seen a million satellite images of this but not a single photo from people on the ground despite the CV having millions of people

The 2015 moment for DJI drones is happening right now for 360 degree, stereoscopic, and VR cameras Every incumbent camera company (Nikon, Canon, Sony, GoPro) is about to go bust. Normal cameras are a perfected technology, with not much room to improve on. VR Cameras are the only path forward, since they require an entire re-design of the camera body Chinese companies like @insta360 @XGRIDS2023 @KandaoVR @DJIGlobal are leading this push. @insta360 and @DJIGlobal are leading the 360 degree camera market. With over 5 million units sold in the last 5 years, and the majority being sold within China. @KandaoVR and @RealSenseai are leading the push for stereoscopic cameras mainly used for robotics @XGRIDS2023 is leading the market for volumetric capture with the new PortalCam and let's not forget about the countless new AR/AI glasses enabling unique POV egocentric capture. While generative AI is generating massive synthetic 2D content, a revitalization and explosion of immersive content is happening now. At braintrance.net, we're betting that all this content needs a host to make sense of everything. Follow this account and @braintrancenet to stay updated on our journey! Lots of big updates to announce soon.