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

GPT-4

openai.com

Post reply on HN
Print view View original

621–630 of 1001 posts

Re: GPT-4

#621

Post by LesZedCB »

jddj wrote: post #571
Imnimo wrote: post #218

A class of problem that GPT-4 appears to still really struggle with is variants of common puzzles. For example: >Suppose I have a cabbage, a goat and a lion, and I need to get them across a river. I have a boat that can only carry myself and a single other item. I am not allowed to leave the cabbage and lion alone together, and I am not allowed to leave the lion and goat alone together. How can I safely get all three…

Silk silk silk silk silk silk. What do cows drink?

haha, brains are truly amazing, both in their resiliency and in their vulnerability. you got me.

Re: GPT-4

#622

Post by DubiousPusher »

Dude said something like "you could hook this up to a calculator". Anyone know if that is implying this generation of model could interface with some kind of symbol processor? Or is he just saying, "in theory", there could be a model that did that?

The math seems much improved and it would be a cool trick if it were emulating a symbol processor under the hood. But humans can do that and we opt for calculators and computers for a reason. IMO, the first most useful thing to come from human-machine interface would be adding a highly reliable Turing machine to your cognition.

If we could do that with one of these models, we could have a pretty strong assumption that long proofs and such were performed with strict rules and the model wasn't falling into some going from the gut, "this equation looks like x" type holes. Which seem like a hazard and make me very uncertain any time I see a paper about using ML to come up with answers to what are essentially logical problems.

Re: GPT-4

#623

Post by raincole »

billythemaniam wrote: post #612
hooande wrote: post #584

After watching the demos I'm convinced that the new context length will have the biggest impact. The ability to dump 32k tokens into a prompt (25,000 words) seems like it will drastically expand the reasoning capability and number of use cases. A doctor can put an entire patient's medical history in the prompt, a lawyer an entire case history, etc. As a professional...why not do this? There's a non-zero chance that i…

If you had full source code that fit into the context, do you think it could reliably answer questions about the code, build unit tests, generate documentation? I ask because that is the software equivalent of what you just described.

All we can do is guessing now until more people get access to the new API. My bet is it can at least generate documentation pretty well.

Re: GPT-4

#624

Post by thecosmicfrog »

Interestingly, ChatGPT has no idea if it's running GPT-3 or GPT-4, since it has no concept of GPT-4 existing in September 2021 (its knowledge cutoff).

Me: Are you running on GPT-3 or GPT-4?

ChatGPT: As of my knowledge cutoff in September 2021, the most advanced version of GPT (Generative Pre-trained Transformer) publicly available was GPT-3, which is the version I am based on. At the time, GPT-3 was the largest and most powerful language model created by OpenAI, with 175 billion parameters.

I am not aware of any public announcement or release of GPT-4 by OpenAI since then, but it is possible that they may have continued to work on new versions of the model in the meantime. However, I have not received any updates on my own capabilities beyond my knowledge cutoff date, so I cannot currently provide information beyond that point.

Re: GPT-4

#625

Post by jahewson »

agnosticmantis wrote: post #605

This is all cute and entertaining, but my digital assistant still remains as dumb as ever and can’t process the simplest of ordinary tasks. I still can’t ask my phone to “add a stop at cvs if it doesn’t add more than 5 minutes to my trip” while driving and using maps/navigation. Is that too much to ask from a superhuman-performing AI that’s mastering all tasks and will disrupt everything? Or maybe the hype is more th…

Just tried this with Apple Maps + Siri and it can do it if the place you’re asking for is not ambiguous but it requires you to press to confirm. It can also show you the amount of time the stop will add in a prompt before hand, but again only visually.

Re: GPT-4

#626

Post by coder543 »

thecosmicfrog wrote: post #624

Interestingly, ChatGPT has no idea if it's running GPT-3 or GPT-4, since it has no concept of GPT-4 existing in September 2021 (its knowledge cutoff). Me: Are you running on GPT-3 or GPT-4? ChatGPT: As of my knowledge cutoff in September 2021, the most advanced version of GPT (Generative Pre-trained Transformer) publicly available was GPT-3, which is the version I am based on. At the time, GPT-3 was the largest and m…

Unless you are using the paid version (ChatGPT Plus), then you are certainly still talking to the GPT-3 model. Even on ChatGPT Plus, my understanding is that you have to specifically select the new model.

Re: GPT-4

#627

Post by zpeti »

mk_stjames wrote: post #577

I just finished reading the 'paper' and I'm astonished that they aren't even publishing the # of parameters or even a vague outline of the architecture changes. It feels like such a slap in the face to all the academic AI researchers that their work is built off over the years, to just say 'yeah we're not telling you how any of this is possible because reasons'. Not even the damned parameter count. Christ.

The sceptical me says its more of a marketing ploy, for people not subscribed to chatgpt pro yet, getting v4 is a good reason.

I wouldn't be suprised if they get millions of new subscribers today.

Re: GPT-4

#628

Post by jackcooper »

DubiousPusher wrote: post #622

Dude said something like "you could hook this up to a calculator". Anyone know if that is implying this generation of model could interface with some kind of symbol processor? Or is he just saying, "in theory", there could be a model that did that? The math seems much improved and it would be a cool trick if it were emulating a symbol processor under the hood. But humans can do that and we opt for calculators and com…

Stephen Wolfram wrote an article [0] about using Wolfarm|Alpha for this you might find interesting.

[0] https://writings.stephenwolfram.com/2023/01/wolframalpha-as-...

Re: GPT-4

#630

Post by jakewins »

d3ckard wrote: post #600
hooande wrote: post #584

After watching the demos I'm convinced that the new context length will have the biggest impact. The ability to dump 32k tokens into a prompt (25,000 words) seems like it will drastically expand the reasoning capability and number of use cases. A doctor can put an entire patient's medical history in the prompt, a lawyer an entire case history, etc. As a professional...why not do this? There's a non-zero chance that i…

I must have missed the part when it started doing anything algorithmically. I thought it’s applied statistics, with all the consequences of that. Still a great achievement and super useful tool, but AGI claims really seem exaggerated.

This paper convinced me LLMs are not just "applied statistics", but learn world models and structure: https://thegradient.pub/othello/

You can look at an LLM trained on Othello moves, and extract from its internal state the current state of the board after each move you tell it. In other words, an LLM trained on only moves, like "E3, D3,.." contains within it a model of a 8x8 board grid and the current state of each square.

Post reply on HN