Public Domain AI https://xitter.cf/t.co/PcBwhip8WA | https://xitter.cf/t.co/2ndEeE3sde
Joined January 2021
- Tweets731
- Following163
- Followers4.8K
- Likes625
Spawning retweeted
More Public Diffusion samples! It’s still very early on (<33% done), but I’m optimistic with the quality of these outputs.
Some interesting examples in 🧵, all prompts in the alt text. My favorites:
ALT A painting of the interior of a cathedral with people sitting on the floor, surrounded by pillars, statues, and stained glass windows. The ceiling is visible at the top of the painting, and the walls are adorned with intricate designs.
ALT A photo of a foggy medieval village with a wide, empty road in the center of the image from the foreground to the background, leading directly into the fog. The ground is made of dark soil and is muddy, with standing pools of water. The ground is heavily disturbed. The village buildings have thatched roofs. There are crudely crafted vendor tables, stands, and carts on the side of the road. No person is present, giving an ominous feeling. Gothic cathedrals can be seen in the far distance. The outline of a hillside is visible in the fog.
Spawning retweeted
Here are some images from our first candidate model for Public Diffusion.
It still needs a lot more time in the oven, but these early results are very promising.
Prompts in the alt text.
ALT "a painting of a snowy mountain landscape with a stream running through it, surrounded by trees, rocks, and a house. The sky is filled with clouds and the mountains are covered in a blanket of snow."
ALT “a painting of a sunset over a body of water with a castle in the distance, surrounded by hills and rocks. The sky is filled with clouds and the sun is setting.”
ALT "an oil painting of a majestic waterfall in the middle of a lush green forest. The waterfall is surrounded by tall trees on either side, and the sky is filled with white, fluffy clouds. The water cascades down the rocks, creating a beautiful and serene atmosphere."
Spawning retweeted
Public Diffusion proves a radical idea: you don’t need infinite data—you need curated data. With just public domain images, the early results rival top-tier AI art models with just 30% training complete. The future isn’t owned. It’s shared. cc: @spawning_ @ncpadge @JordanCMeyer
Spawning retweeted
First visual model trained on fully open data coming. Like for text, the striking thing is not only that it is possible in the first place but that it brings so much more variety and diversity of outputs. There’s an entire world beyond web archives.
Spawning retweeted
███░░░░░░░░░ ~25% trained
"A painting of a mountain lake with a boat in the foreground, surrounded by lush green grass, trees, and rocks. The sky is filled with white, fluffy clouds, creating a peaceful atmosphere."
Join the private beta to help test Public Diffusion
@spawning_
PD is a foundation image model trained on 30m highly aesthetic images from the public domain
We will show it is possible to have excellent AI models for art that do not require training on copyrighted material
Link below
Spawning retweeted
While lawsuits about training data and copyright law wend their way through the courts...
...@spawning_ is building image generation tools trained on public domain/CC0.
Check out Public Diffusion.
Join the private beta to help test Public Diffusion
@spawning_
PD is a foundation image model trained on 30m highly aesthetic images from the public domain
We will show it is possible to have excellent AI models for art that do not require training on copyrighted material
Link below
Spawning retweeted
What if AI image models weren't stealing from artists?
@matdryhurst joined us to talk about @spawning_ - Spawning just released PD12M, a dataset containing over 30 Terabytes of public domain image and text data for AI models 👀 🌱
1/5 Mat explains why this is such a big deal 👇
Spawning retweeted
Public AI models are possible!
Spawning just released the largest public domain image dataset
Now we are training Public Diffusion, a public domain image model with no IP concerns
This will help any artist fine tune, and own, their own models on their own terms
DIY
We’ve put 12.4 million of the most beautiful images in the public domain into an image-text dataset with community governance.
Public Domain 12M is completely free. Its 30+ terabytes of images are hosted on AWS, so downloading them is fast and won’t negatively impact the cultural heritage organizations who originally shared them.
Spawning retweeted
Our colleague @cullenmiller comprehensively breaks down the recently resolved LAION case and what it means for AI and datasets
The recent German court ruling on the LAION dataset has far-reaching implications that could impact your work in AI, data governance, and copyright regulation.
In our latest analysis, @cullenmiller explores:
- Why the court's distinction between dataset creation and model training matters more than ever
- How this ruling aligns with the EU AI Act
- Why your current data governance tools might be outdated
- What 'time-of-training' checks mean for your IP or training pipeline
Whether you're training models, protecting intellectual property, or shaping AI policy, we break down our read on the ruling—seeing it as a clarifying development that is in line with Spawning's methodology and tools for 'time-of-training' opt-out checks.
spawning.substack.com/p/refl…
Spawning retweeted
recommended read from @spawning_
The recent German court ruling on the LAION dataset has far-reaching implications that could impact your work in AI, data governance, and copyright regulation.
In our latest analysis, @cullenmiller explores:
- Why the court's distinction between dataset creation and model training matters more than ever
- How this ruling aligns with the EU AI Act
- Why your current data governance tools might be outdated
- What 'time-of-training' checks mean for your IP or training pipeline
Whether you're training models, protecting intellectual property, or shaping AI policy, we break down our read on the ruling—seeing it as a clarifying development that is in line with Spawning's methodology and tools for 'time-of-training' opt-out checks.
spawning.substack.com/p/refl…
Spawning retweeted
Sind Datasets für das KI-Training machbar, deren Inhalte urheberrechtsfrei sind und/oder von den Urhebern bewusst und freiwillig zur Verfügung gestellt werden? Dass das möglich ist, will @spawning_ mit SourcePlus beweisen. 1e9.community/t/ki-training-…
Spawning retweeted
Great piece where @kevinroose covers @medialab research showing that the web is balkanizing, as we predicted
@spawning_ built our ai.txt upgrade on robots.txt to try and avoid this
Ai.txt distinguishes between crawling and scraping, keeping the web open for the many rather than shutting it down for everyone in avoidance of the few
nytimes.com/2024/07/19/techn…
Spawning retweeted
✨New Preprint ✨ How are shifting norms on the web impacting AI?
We find:
📉 A rapid decline in the consenting data commons (the web)
⚖️ Differing access to data by company, due to crawling restrictions (e.g.🔻26% OpenAI, 🔻13% Anthropic)
⛔️ Robots.txt preference protocols are ineffective
These precipitous changes will impact the availability and scaling laws for AI data, affecting coporate developers, but also non-profit and academic research.
🔗 dataprovenance.org/consent-i…
1/
Spawning retweeted
.@spawning_ recently launched a new project, Source.Plus:
“Our goal is to make it easy for rights holders to offer their media for use in generative AI training — on their own terms — and frictionless for developers to incorporate that media into their training workflows," shared CEO and Co-Founder @JordanCMeyer.
Learn more in @TechCrunch from @Kyle_L_Wiggers: techcrunch.com/2024/06/11/sp… #ethicalAI #AItraining
Spawning retweeted
Cool thing less people know is: SD3 was trained using opt-out framework @spawning_ - and it does mean a huge step towards human artists and away from "AI steals things" :D Spawning - and later Source.Plus are the future!
Spawning retweeted
I enjoyed speaking with @ezraklein about AI, art, data and @spawning_
I think it is more accurate to describe current systems as Collective Intelligence
This shifts the emphasis from the improbable claim that models have a mind of their own towards the human engineers and data contributors who have created systems greater than the sum of their parts
What does the future of A.I. art look like? Sound like? Feel like? In this episode of The Ezra Klein Show, @hollyherndon joins @EzraKlein to discuss “data manners” and the weird and wonderful potential of collective creation. nyti.ms/3WTJRQD
Spawning retweeted
Opt in image datasets are here!
Early access to Source.Plus
A platform to curate, enrich and download non-infringing data for model training
We launch with 200 TB of high quality public domain image data, EU compliant, and more than enough to train a SoTA base model
Just the beginning
Spawning retweeted
If we can have clean energy, we can have clean training data. Love this from @spawning_ xitter.cf/matdryhurst/status/179…
Early access to Source.Plus
A platform to curate, enrich and download non-infringing data for model training
We launch with 200 TB of high quality public domain image data, EU compliant, and more than enough to train a SoTA base model
Just the beginning
Spawning retweeted
.@spawning_ doing god's work
Early access to Source.Plus
A platform to curate, enrich and download non-infringing data for model training
We launch with 200 TB of high quality public domain image data, EU compliant, and more than enough to train a SoTA base model
Just the beginning