For the first #Botober collection I generated, I had GPT-3 generate "things". Here's an alternate collection in which I had the neural net generate "concepts". It's different, but I'm not sure if it's easier.
I'm hand-curating the GPT-3 generated #Botober prompts because many of the neural net's drawing prompts are terrible for one reason or another.
Here are some of the prompts that were unfairly difficult.
The neural net suggested #Botober drawing prompts based on how the words fit, and not on any concept of how the objects might look. This list makes that abundantly clear.
There are several ways of getting GPT-3 to generate drawing prompts.
For #Botober I chose to give it a short story to complete, in which it had to generate lines for an erratic station AI. The neural net took over after "1."
@janellecshane I felt the urge to sing all of these, but I did so under my breath, so that my family wouldn't think I was even weirder than usual.
@janellecshane pile of logs more popular than expected!
Worms pulsing in a black light is really wow!
@janellecshane stupid qu? : do you think it can fill in "call for project" dossiers in art? that would be a relief to apply to numerous calls while not wasting time scribbling stupid txts and not creating ;)
@janellecshane ho good lord of inculture : probably not! ... given the ratio of word/picture/boxes to fill I went accross those years, it should really help then even without knowing anything about pictures (and we can generate them from samples + text titles probably with other gen algo ;p )
... (feeling some reality drift here.... )
@janellecshane I had fun reading them, but I'm suspicious. How can they be so funny? Also "Juice for humans" comes up in two otherwise completely different runs? And both times it's the seventh entry. Were the runs correlated somehow? From what I understand about GPT-3, which is a vast network, that's a remarkable repetition considering the difference in prompts (story vs list). If I'm mistaken, GPT-3 and I have the same humor.
@mplammers I'm curating the best prompts that GPT-3 generates - the screenshot that has the story in it is an example of raw output from GPT-3. I did end up using #7 from that GPT-3 run for one of the prompt lists. I tried where possible to keep the numbers the same (although if I had two #7s I liked, I'd change one to #8 or another slot I had open). Hope this helps! I did see "giraffe" repeated a few times.
@janellecshane Sure helps. I think it's a remarkable display of synergy between writer and computer and as such I find it refreshing and new. I do maintain (but that is my personal opinion) that presenting more openly your procedure, curation and your part in this wonderful synergy would only make it better. Thank you for sharing these!
@Sandra The second one, "concepts" is as close to final as it gets. But people are free to do whichever list they like best, or to mix and match.
The Wandering Shop is a Mastodon instance initially geared for the science fiction and fantasy community but open to anyone. We want our 'local' timeline to have the feel of a coffee shop at a good convention: tables full of friendly conversation on a wide variety of topics. We welcome everyone who wants to participate, so long as you're willing to abide by our code of conduct.