Jordan Meyer
@JordanCMeyer
Followers
1K
Following
1K
Statuses
806
Cofounder/CEO @spawning_ | Zillow Prize winner | pfp by @irisluckhaus
Chapel Hill, NC
Joined July 2010
RT @matdryhurst: It’s surreal that I be the subject of controversy over the provenance of data used in my art (about the provenance of data…
0
13
0
@SearchTrut69841 We're working on fine-tuning with the first set of beta testers through February. I think a full release won't happen until early summer, but if all goes well, the beta should grow before then.
0
0
2
RT @hollyherndon: Happy AI breakthrough day! 🎄 I'm so proud this beautiful book is finally out! This was a huge group effort and covers…
0
26
0
RT @matdryhurst: Interview with @JordanCMeyer and @ncpadge who are building Public Diffusion, with some more beautiful examples from the mo…
0
1
0
RT @spawning_: I sat down with CEO Jordan Meyer and CTO Nick Padgett to talk all things Public Diffusion, from fine-tuning, to prompting li…
0
2
0
RT @ShayneRedford: ✨New Report✨ Our data ecosystem audit across text, speech, and video (✏️,📢,📽️) finds: 📈 Rising reliance on web, synthet…
0
42
0
@v4f4t4 @micha @spawning_ If there's an ensemble model that can route and blend the more specific fine-tunes, that approach could also route compensation to the people whose work was used to create the specialist models.
0
0
3
RT @matdryhurst: The king of YouTube AI image tutorials covers Public Diffusion ❤️ Thanks Olivio. To confirm the weights will be unconditi…
0
2
0
RT @ncpadge: The model also seems to be good at differentiating between mediums. These images used the same prompt but changed the medium.…
0
1
0
@Peterthe_roman Thank you! I agree img2img will be a huge boost. Especially given that the model has (imo) great fidelity, but its knowledge of the world is limited by the PD dataset.
1
0
2
@meatcrowning Sure thing! With the Lumina-next model, we'd need to do 2x the amount of training to get to a 512px model that we had to do for this one. We're testing quite a few new ideas, so we felt like any potential AE quality hit was worth taking since this is likely to be a placeholder.
0
0
1
RT @ncpadge: More Public Diffusion samples! It’s still very early on (<33% done), but I’m optimistic with the quality of these outputs. So…
0
8
0
@DarrylMason @spawning_ I've been so focused on the model training that I haven't yet, but I celebrated public domain day for the first time last year and we should probably make it a @spawning_ holiday!
0
0
3
@larimus23 @spawning_ The first round of beta invites will go out next week. We'll need to keep it small at first, but we'll do our best to open it up quickly. The beta will run for several months while we try out models with interested artists. Then we'll train a full model and release.
0
0
2
@EvanBugmann @ecutruin @spawning_ The final training run (after we run a pretty long beta to figure out which model architectures will be the best fit for a model like this) will be open weights with commercially permissive licensing. I think of LoRAs as the primary use for this, given the public domain dataset.
0
0
1
@woctordho I'm expecting that anime will need some fine tuning, but I agree it will be interesting to find out if I'm wrong about that!
0
0
0