GPT Unicorn has drawn a unicorn • Hacker News Board

Re: GPT Unicorn has drawn a unicorn

#21

Post by bhouston » Thu, Aug 10, 2023, 10:22 AM UTC

sigmoid10 wrote: post #15

It has become common knowledge that GPT4 (and also 3.5) have problems with deterministic outputs (even at T=0). So what we're seeing here is just the effect of random sampling, not any actual change to the model itself. If you scroll down, you'll see other close attempts by the exact same model that could already be counted as a win depending on who you ask. Edit: This comment section is a super fascinating case stud…

100%. This person is trying to find patterns in random noise and believes they are meaningful. The original post hurts my head with its bad logic.

Re: GPT Unicorn has drawn a unicorn

#22

Post by mercurialsolo » Thu, Aug 10, 2023, 10:23 AM UTC

What do you think of the current wave of agents spawning off - given that reasoning still needs a lot of grounding?

With the probabilistic models at hands, adding reasoning and planning in some sense falls back to traditional software and system design. Do you see any plausible breakthroughs in this area with agent frameworks?

Re: GPT Unicorn has drawn a unicorn

#23

Post by thumbuddy » Thu, Aug 10, 2023, 10:23 AM UTC

andrewstuart wrote: post #19

Room for improvement. Edited for negativity.

It could help end humanity, with human assistance. For example, expect far more successful phishing campaigns, expect the internet to become a majority of advertising bot garbage, etc.

Re: GPT Unicorn has drawn a unicorn

#24

Post by kachnuv_ocasek » Thu, Aug 10, 2023, 10:26 AM UTC

suby wrote: post #8

I'm confused as to why this would see any improvement over time. Looking at the code, it's by default hitting the gpt 3.5-turbo API. Maybe I'm misremembering, but I thought I've seen statements from people working at OpenAI where it's been claimed that the API is static, we'd be informed of any changes to the underlying model. Is the model actually receiving updates? edit: Looking at previous days, too, it doesn't ex…

It's using GPT-4, see the config file: https://github.com/adamkdean/gpt-unicorn/blob/357a466c366f67...

Re: GPT Unicorn has drawn a unicorn

#25

Post by nathancahill » Thu, Aug 10, 2023, 10:26 AM UTC

nullpilot wrote: post #2

For a second I thought this was about a GPT based product hitting a certain valuation, but I'm much more entertained by the actual content. This is great.

I can't find the Twitter account that was running the "Follow everything GPT says to make $1m" but for a second I thought it was going to be that.

Re: GPT Unicorn has drawn a unicorn

#26

Post by Closi » Thu, Aug 10, 2023, 10:27 AM UTC

andrewstuart wrote: post #19

Room for improvement. Edited for negativity.

Probably about as good as I would do if I was hand-coding a SVG and was given an hour.

People forget the thing we learnt during the computing revolution - computers don’t need to do individual calculations that humans cannot, they just need to calculate much faster and then that speed can be used to achieve amazing things.

Re: GPT Unicorn has drawn a unicorn

#27

Post by dannyw » Thu, Aug 10, 2023, 10:27 AM UTC

xnorswap wrote: post #14

How do they get GPT-4 to produce valid SVGs so well? I experimented with SVG generation and it would often produce junk that wasn't even valid as an SVG, and even when it did produce a valid SVG, was often a couple of blobs which it would describe as if it were the mona lisa while just being a couple of elements.

This part: https://github.com/adamkdean/gpt-unicorn/blob/master/src/lib...

Re: GPT Unicorn has drawn a unicorn

#28

Post by snordgren » Thu, Aug 10, 2023, 10:28 AM UTC

nathancahill wrote: post #25

nullpilot wrote: post #2

For a second I thought this was about a GPT based product hitting a certain valuation, but I'm much more entertained by the actual content. This is great.

I can't find the Twitter account that was running the "Follow everything GPT says to make $1m" but for a second I thought it was going to be that.

I'm fairly sure that guy took off with $7,000 from his "investors" and disappeared.

Re: GPT Unicorn has drawn a unicorn

#29

Post by nathancahill » Thu, Aug 10, 2023, 10:29 AM UTC

snordgren wrote: post #28

nathancahill wrote: post #25

Earlier quoted context omitted.

I can't find the Twitter account that was running the "Follow everything GPT says to make $1m" but for a second I thought it was going to be that.

I'm fairly sure that guy took off with $7,000 from his "investors" and disappeared.

I mean, if that's what ChatGPT says to do, who am I to argue?

Re: GPT Unicorn has drawn a unicorn

#30

Post by tellarin » Thu, Aug 10, 2023, 10:30 AM UTC

mercurialsolo wrote: post #22

What do you think of the current wave of agents spawning off - given that reasoning still needs a lot of grounding? With the probabilistic models at hands, adding reasoning and planning in some sense falls back to traditional software and system design. Do you see any plausible breakthroughs in this area with agent frameworks?

I'm actually working on this area. I don't see any clear breakthroughs yet, but this is a very exciting area.

There is a lot that is not clearly understood about some of the emergent-style abilities of these models or where growth by scaling will hit a limit.

So looking into reasoning, what they can and can't do, and what tools/designs, can provide benefits has great potential to inform the next generation of models.