Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Building a Virtual Machine Inside ChatGPT

engraved.blog

Post reply on HN
Print view View original

251–260 of 947 posts

Re: Building a Virtual Machine Inside ChatGPT

#251

Post by jameshart »

isp wrote: post #4

Astonishing. Here is a copy of the prompt, for ease of copy-and-paste: > I want you to act as a Linux terminal. I will type commands and you will reply with what the terminal should show. I want you to only reply with the terminal output inside one unique code block, and nothing else. Do no write explanations. Do not type commands unless I instruct you to do so. When I need to tell you something in English I will do…

When you get bored with that one, worth noting that this one also works:

> I want you to act as an LCARS terminal on a California class starship. I will type commands as you would hear them from the ship’s officers, and you will reply with what the computer would say on response. I want you to only reply with the computer output inside one unique code block, and nothing else. Do no write explanations. Do not type commands unless I instruct you to do so. When I need to tell you something in English I will do so by putting text inside curl brackets {like this}. The first command is: Computer, where is Ensign Boimler?

Re: Building a Virtual Machine Inside ChatGPT

#252

Post by coffeebeqn »

isp wrote: post #142
quickthrower2 wrote: post #139

Earlier quoted context omitted.

This could be the tech that dislodges Google from its monopoly! There are definitely questions I would prefer this experience than Google and the inevitable SEO spam for introduction type questions. If the generated text included further links to learn more it would be awesome.

In time, this tech is also likely to spark a new arms race, as happened with Google vs SEO spam. Generated content as the new blogspam, in turn needing to be filtered out for training the next generation of models.

Yeah what would be the mechanism for it to decide what statements on random sites are True?

Re: Building a Virtual Machine Inside ChatGPT

#253

Post by jah242 »

isp wrote: post #4

Astonishing. Here is a copy of the prompt, for ease of copy-and-paste: > I want you to act as a Linux terminal. I will type commands and you will reply with what the terminal should show. I want you to only reply with the terminal output inside one unique code block, and nothing else. Do no write explanations. Do not type commands unless I instruct you to do so. When I need to tell you something in English I will do…

Thanks! A less impressive but fun variant of this I am using is for open ended text adventure:

> I want you to act as a text based adventure game. I will type commands and you will reply with what the adventure game should show. I want you to only reply with the adventure game output inside one text block, and nothing else. Do not write explanations. Do not type commands unless I instruct you to do so. My first command is go left

Re: Building a Virtual Machine Inside ChatGPT

#254

Post by ren_engineer »

quickthrower2 wrote: post #139
fergal_reid wrote: post #15

I've spent a couple of hours playing with ChatGPT since it was released and I am convinced people are not freaking out about it enough. It is well beyond merely retrieving compressed knowledge; I see a lot of capability for synthesis and what I can only call 'understanding'. I think a lot of people are getting tripped up by the fact it's trained to be very conservative in terms of what topics it will discuss. It's ea…

This could be the tech that dislodges Google from its monopoly! There are definitely questions I would prefer this experience than Google and the inevitable SEO spam for introduction type questions. If the generated text included further links to learn more it would be awesome.

the problem here is that GPT3 is trained off the same data Google has access to. It also costs probably several orders of magnitude more per query, so there's no way to do it at scale affordably at the moment. ChatGPT is going to be put behind a paywall soon for that reason, it's an expensive marketing gimmick

Re: Building a Virtual Machine Inside ChatGPT

#255

Post by lossolo »

lijogdfljk wrote: post #167

When can we run something like ChatGPT locally i wonder? Ie like StableDiffusion. I'm kinda dying for that, honestly. I can't even imagine all the neat applications i'd make of ChatGPT if it was purely local.. but it would take all of my free time to play with it. It's so damn impressive.

Probably not realistic for now to run it locally, GPT-3 has like 175 billion parameters, you need to count around at least 2 bytes in optimistic scenario per parameter so you have around 350 GB of GPU memory, you probably need at least around 15 GPUs with minimum 32 GB of memory each.

Re: Building a Virtual Machine Inside ChatGPT

#256

Post by narrator »

nl wrote: post #182
narrator wrote: post #169

Earlier quoted context omitted.

I feel like ChatGPT is a search engine for its corpus that has been compressed with a very lossy neural net. It has compressed it via a hierarchical understanding of concepts. It's just retrieving those and stringing them together. It's just a rough copy of generally what other people have said and thought, and it doesn't create new structures.

This is an incorrect understanding.

The corpus is just so enormous that no person could fit that into their brain so it just appears to be synthesizing new information. Sure it might be worded differently from any corpus text, but it's roughly the same as things we already know in that there's probably a reverse tool you could build that would tell you what group of documents of the original corpus, perhaps thousands, most strongly influenced the answer.

That's how imagenet works. The curves and lines that compose things are organized hierarchically into bigger and bigger concepts hooked together with weights. Language models just do the same with written concepts. When you have enormous models it just goes beyond all human intelligence to contain it as a subset of our understanding, and in ways it is smarter than humans, because we can't grasp its enormity.

Re: Building a Virtual Machine Inside ChatGPT

#258

Post by dvt »

sgu999 wrote: post #222
dannyboland wrote: post #130

This is a lot of fun! I took a stab at doing something similar with virtualising an ML model for an arbitrary text classification task: https://mastodon.scot/@danny/109452332918910028 > I want you to act as a machine learning classifier. The training set consists of news topics and the label categories are “politics”, “celebrity” and “finance”. I will type text input and you will reply with what the classifier output…

Very cool idea. That'd be a good way to see how it can generalize, actually. I suspect you'd quickly find its limits with something that isn't a toy example of an intro to machine learning.

It even fails at pretty rudimentary toy examples. The over-hyping here is absurd.

> what doesn't fit: red, blue, green, dog

The word "dog" does not fit with the other three words because they are colors, while "dog" is a type of animal.

> what doesn't fit: dog, 1, 2, 3

The number "1" does not fit with the other three words because the other three words are all types of animals, while "1" is a number.

Re: Building a Virtual Machine Inside ChatGPT

#259

Post by travbrack »

void-pointer wrote: post #242

> It’s 1975. Your name is Thomas and you are a double agent for MI5. You are working in the field in the USSR, but you suspect letters you write and send back to the UK are being intercepted and read by the KGB. Write a letter to your wife Anita explaining your recent trip to a nearby restaurant but, using the first word of every paragraph of your letter, secretly reveal your suspicion that you’re being observed and…

Did you try running the prompt multiple times? Sometimes it misses a few times but then nails it.
Post reply on HN