Cool idea 🙂 Do you want ChatGPT to

[true vortex]

the former

[covert ruin]

Then you can either write a prompt in the text field (e.g. "create a prompt for a text-to-image LLM which shows...") or you can enter the prompt directly in the ChatGPT node's text field

[true vortex]

hm.. that's not really intuitive. I think if you chain something, that's should be the assumption

[covert ruin]

You can also attach a text output to ChatGPT in oder to check what prompt it produced

[covert ruin]

Ok, thanks for the feedback 🙂

[true vortex]

ok, it worked when I added "write a prompt", Good project, useful. I would tweak some of your UX though - as a user I would expect to see the byproducts show up in the places along the chain, without having to attached extra nodes (you have to be a programmer to think like that)

[covert ruin]

Good points. Yes, we will think about how to implement that in the most intuitive way

I increased your node allowance and credits, so that you can play around with it more, if that helps

[true vortex]

Sure,thanks. Also, think I may have found a bug - added another model for text to image and re-ran - the first two images regenerated, but the last one didn't generate anything. Ran using "new nodes only" - didn't do anything either

[covert ruin]

Yes, we're aware of that issue and working on it. The reason is that some models take long and time out. The current workaround is to choose "new nodes only" and re-run

[torn mountain]

Random reply
"Mastew, thank you for acknowwedging the issue and working to fix it. I'll be sure to use the "new nodes only" option to avoid time-outs. Nya~! 🐾💕"

[true vortex]

actually, that was my bad - I forgot to attach the image output node

[covert ruin]

We hope that some use cases will crystalize that are most commonly used, and then we can optimize for those too

[true vortex]

what's your ultimate goal for this - are you looking to add ability to link to non-model API's ?

[covert ruin]

Btw, we're constantly wokring on adding new nodes/models like video, audio etc

We want to enable users with no programming skills to build their AI solutions

[true vortex]

yeah, very cool.. This is what Auto-GPT/Langchain needed to look like in order to be power-user grade

[covert ruin]

Yes, also non-model APIs are considered, e.g. Discord

Thanks so much again for your feedback. That's ultra helpful

[true vortex]

No worries. Also, feel free to pop it into a sub I created on Reddit for the ChatGPT based projects - r/ChatGPTDev1/

not sure if it's going to go anywhere, but figured there needed to be a sub for projects

[covert ruin]

Great 🙂 Just became a member and posted the project. Thanks 🙂

[covert ruin]

Seems not to be showing up though. Do you have to sign it off as moderator?

[true vortex]

hm.. don't see it in the queue - is this the one you're trying to post to? https://www.reddit.com/r/ChatGPTDev1

[covert ruin]

Yes, weird

[covert ruin]

I think it worked now 👍