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

GPTs: Custom versions of ChatGPT

openai.com

Post reply on HN
Print view View original

231–240 of 341 posts

Re: GPTs: Custom versions of ChatGPT

#231

Post by garbanz0 »

When I saw this, I figured it was a clear step back from simply using plugins in the main ChatGPT view. It's basically plugins, but with extra prompting and you can only use one at a time.

But if you look at projects like Autogen ( https://github.com/microsoft/autogen ), you see one master agent coordinating other agents which have a narrower scope. But you have to create the prompts for the agents yourself.

This GPTs setup will crowd-source a ton of data on creating agents which serve a single task. Then they can train a model that's very good at creating other agents. Altman nods to this immediately after the GPTs feature is shown, repeating that OpenAI does not train on API usage.

Prediction: next year's dev day, which Altman hints will make today's conference look "quaint" by comparison, will basically be an app built around the autogen concept, but which you can spin up using very simple prompts to complete very complex tasks. Probably on top of a mixture of GPT 4 & 5.

Re: GPTs: Custom versions of ChatGPT

#233

Post by Racing0461 »

singularity2001 wrote: post #165
Racing0461 wrote: post #100

Earlier quoted context omitted.

I doublt it's fine tuning (actually changing the model weights). It's more like "im going to paste a text blob then in the following chats i will ask questions about it) type inner prompt.

For longer documents it uses vector embeddings

How's that different from pasting the text in the first chat and running the vector embedding step on the text on the server (maybe at least bypassing the chat text limit)? Does this fix the amnesia issue where the info from chats longer than the context length is forgotton because the document isn't baked directly into the weights like fine tuning?

Re: GPTs: Custom versions of ChatGPT

#235

Post by omarfarooq »

tsunamifury wrote: post #203
Sai_ wrote: post #186

Earlier quoted context omitted.

Isn’t that the crux of the conversation? Jasper had to fight to survive because they were a wrapper.

What isn't a wrapper? Honestly everyone turns they nose in the air and claims all superiority bit I challenge you to find a use case that isn't just a 'wrapper'

Not all wrappers are created equal.

Re: GPTs: Custom versions of ChatGPT

#236

Post by BoorishBears »

Sai_ wrote: post #186
BoorishBears wrote: post #138

Earlier quoted context omitted.

... is what people who don't understand positioning will parrot time and time again. Jasper isn't having a good time, but you'd think the fact anyone can produce better output than they did after spending millions of dollars in GPT-3 based pipelines for $20 a month would mean they're dead dead. But instead they went and changed their positioning, changed who their target market is, adjusted the UX, the messaging, and…

Isn’t that the crux of the conversation? Jasper had to fight to survive because they were a wrapper.

That's the complete opposite of reality: Jasper could only fight to survive because they were a wrapper.

Imagine if Jasper had raised and built their own GPT-3 alternative from the ground up at 100x the cost targeted at writing: it wouldn't have generalized like OpenAI's models given the different goals in training, they'd have spent orders of magnitude more, they'd currently be scrambling to partner with some non-OpenAI provider to play the role OpenAI does for them today...

Jasper "only" had to start integrating new APIs to keep up with the SOTA in quality, and instead of burning precious runway on R&D they get to focus on rebranding and driving home that positioning.

Re: GPTs: Custom versions of ChatGPT

#237

Post by LouisvilleGeek »

garbanz0 wrote: post #231

When I saw this, I figured it was a clear step back from simply using plugins in the main ChatGPT view. It's basically plugins, but with extra prompting and you can only use one at a time. But if you look at projects like Autogen ( https://github.com/microsoft/autogen ), you see one master agent coordinating other agents which have a narrower scope. But you have to create the prompts for the agents yourself. This GPT…

Curious if anyone that has access can verify that we can still build "actions" via localhost like we previously did with plugins?

Re: GPTs: Custom versions of ChatGPT

#238

Post by shmoogy »

nextworddev wrote: post #208
minimaxir wrote: post #23

The GPT Store will prove to be an interesting moderation and quality control experiment for OpenAI. Apple/Google have spent a lot of time and money on both of those things and they still have issues, and that's not even accounting for the fact that AI growth hackers will be the primary creators of GPTs. And a revenue sharing agreement will provide even more incentive to do the traditional App Store marketing shennani…

I wouldn't be surprised if most of the moderation is done via elaborate AI automation. Also RIP chatgpt plugnins

Not sure about currently - but budget increases seemed to be manually reviewed in a queue. I had to increase monthly budget amounts twice and it took like 10-30 hours for each approval, interesting to see how their processes advance.

Re: GPTs: Custom versions of ChatGPT

#239

Post by drusepth »

imdsm wrote: post #76
truakon89 wrote: post #13

But where is the option/link to create a GPT? I can't find it

1 pm PST — path /create

It's long after 1pm PST -- does anyone else have access to this yet? I'm on a paid US account (so I assume I should have access) but still get "You do not currently have access to this feature".

Maybe it's because I'm out of the country today?

Re: GPTs: Custom versions of ChatGPT

#240

Post by bertil »

What is your impression of how respectful of their ecosystem this was? I saw many “OpenAI killed my start-up”-type reactions. Still, it feels like they are trying to handle obvious use cases in the ecosystem and make things easier to integrate canonically and more transparently. And to handle legal concerns. It was their first ecosystem conference.

Then again, iOS felt like a very promising ecosystem when the iPhone 4 came out; I feel like it turned into an addendum/duplicate of the web rather than something genuinely original. Android seems more vivid, but not much more.

Post reply on HN