valine

38 ms·

481.
by valine 4y ago
It’s an intensionally misstated problem to see if gpt4 will correctly take the lion across first and leave the goat.
482.
by valine 4y ago
For anyone wondering you bring the lion across. The trick is that it’s the lion that eats the cabbage not the goat.
483.
by valine 4y ago
Where are you getting the 1T parameter number? Is that from the leaks/speculation or is that an official number from OpenAI?
484.
by valine 4y ago
Maybe evil is mothballing technology that could benefit society because of potential impacts to your ad business.
485.
by valine 4y ago
Reminds me of the google april fools prank in 2009 where they announced gmail would be able to read and respond to emails for you. http://hoaxes.org/af_database/permalink/gmail_autopilot
486.
by valine 4y ago
I don’t see any problem with an abundance of coding prompts. The type of person who runs a local LLM is probably using it for coding anyway.
487.
by valine 4y ago
7B parameters is next to nothing when compared to gpt3. If 7B works as well as it does here, A fine tuned 65B model could very easily achieve chatGPT level performance.
488.
by valine 4y ago
It can look like anything. Sometimes it will answer your questions, other times it will continue the question like its the one asking. I've also seen it randomly output footers and copyright like it just got to the end of a webpage. It
489.
by valine 4y ago
FW locks aren’t effective. Most of those locked cards have jailbreaks to allow full speed crypto mining.
490.
by valine 4y ago
Our saving grace seems to be the insatiable push by the gaming industry for better graphics at higher resolutions. Their vision for real-time path traced graphics can’t happen without considerable ML horsepower on consumer level graphics ca
491.
by valine 4y ago
What about magnetic confinement fusion? My understanding was that tokamaks use superconductors to produce their magnetic fields. Does a low temperature superconductor help with that?
492.
by valine 4y ago
Nvidia A100 is exactly that. It has lower cuda performance than a RTX 4090, and is almost entirely geared toward ML workloads.
493.
by valine 4y ago
It’s clickbaity because it invites the reader to anthropomorphize ChatGPT.
494.
by valine 4y ago
Computers and the internet looked like bubbles until they didn’t. Not every craze is the next big thing, but for LLMs it’s still too soon to say anything with certainty. The people comparing LLMs to autocomplete have no more legitimacy than
495.
by valine 4y ago
I like to think of stable diffusion as decompressing a text prompt into a 512x512 image. In some sense the 10 word text prompt and seed contains the same amount of information as a 262k pixel image. You could say to a transformer a picture
496.
by valine 4y ago
Black holes probably don’t have infinite density but that doesn’t stop us from finding utility in the theory of relativity. Even if pixels are not little squares it’s a useful abstraction.
497.
by valine 4y ago
Human curated AI generation will soon be indistinguishable from human made content and no less useful. A human’s stamp of approval is what gives content value, not the labor of creating it.
498.
by valine 4y ago
Blender is the counter example. Usage of Blender far exceeds the close source alternative.
499.
by valine 4y ago
I installed the extension and asked it to reply to your comment, and thought the response was impressive. I myself am concerned about the privacy implications. "Hi yunwal, Regarding your concern about the privacy of Dust XP1, it's
500.
by valine 4y ago
I think the issue is that the people building wood apartments aren’t the same people who care enough to ensure there are no small holes in the wall.
501.
by valine 4y ago
RLHF models like ChatGPT seem to be better at listening to instructions than your standard LLM.
502.
by valine 4y ago
Mathematicians are more than just number crunchers. Perhaps the most important part of a mathematicians job is formally defining problems. Even if the next generation of ChatGPT can solve any math problem you give it satisfactory, you still
503.
by valine 4y ago
The 3DS don’t do head tracking, it was a real 3D display that displayed a unique image to each of your eyes.
504.
by valine 4y ago
Amazon built an entire phone around the idea back in 2014. They had cameras on each corner of the phone to improve the effect. https://www.theverge.com/2014/6/18/5821294/amazons-fire-phon...
505.
by valine 4y ago
I’m imagining the AI makes tools it can use by itself. My entire premise depends on you having enough money to fund some AI tinkering, a few Arduinos and servos seems like fair game.
506.
by valine 4y ago
Presumably running the lathe/printer, drawing schematics and providing step by step directions unique to the materials you can afford.
507.
by valine 4y ago
I bet with a lathe, a 3d printer and AGI you could make the tools you need to build an engine. I’m not sure why you would though, since electric motors are both the future and 3D printable.
508.
by valine 4y ago
I don’t know about a factory, but 3D printers, lathes and Arduinos are pretty cheap. Even PCBs are manufacturable at home with a little creativity and patience. If you’ve ever watched Stuff Made Here on youtube, imagine people inventing thi
509.
by valine 4y ago
Can you think of any category of software that doesn't eventually have an open source alternative? There are many open source projects orders of magnitude more complicated than a LLM. The linux kernel, Blender, Firefox, etc. These LLMs
510.
by valine 4y ago
If the software is so dangerous where are all the crashes? FSD beta is running on 400k+ vehicles. I’ve got it running on my car and my biggest complaint is that the software is too hesitant. You can complain about Tesla’s methodology all yo

More ›