//=time() ?>
Woot! The 1st checkpoint of openai's model, tuned on the comics faces dataset is available for early access.
#aiart #discodiffusion #comics
https://t.co/oGFRC2wOy3
【Quantum Kernel Machine Learning】
The general task
of machine learning
is to find
and study patterns
in data.
For many datasets,
the datapoints
are better understood
in a higher dimensional feature space,
through the use
of a kernel function
Done! Pushed into a rgb branch, in case things go awry. If anyone has a good RGBA dataset to help me test this out I can then merge it into main (@duskvirkus
@dvsch @makeitrad1) https://t.co/UJdIb8xHzu
Images can be generated as of now, like this one with a scratch RGBA network https://t.co/Sw6AREO1DU
New @LINZLDS #lidar point cloud dataset covering ~165 sq km of the Taranaki region in New Zealand now available on OpenTopography. Data were captured in 2021 for the Taranaki Regional Council @TaranakiRC by AAM Ltd: https://t.co/cVgnMfGacC
I am taking help from fellow #actionfigure collectors for #thistoydoesnotexist, with the main thing I need (besides $ being HQ figure pics. I made this guide to show what I"m looking for, and for examples of stuff I've approved, the dataset is here: https://t.co/ma7UicWEF7
【MNIST dataset】
The MNIST dataset is
an acronym
that stands
for the Modified National Institute
of Standards
and Technology dataset.
It is a dataset
of 60,000 small square
28×28 pixel grayscale images
of handwritten single digits
between 0 and 9.
#QuantumComputing
OpenCLIP released a new pre-trained ViT-B/16 model on the LAION-400M dataset! 🥳
Link: https://t.co/cMVef53Gl5
"A mecha robot in a favela by James Gurney"
Disco Diffusion 5.1 - OpenCLIP ViT-B/16 LAION-400M
@AVShonenkov The additional effects sound really amazing!
Btw I hooked nev's colab with the newest Surrealist XL dataset and even without any further optimization it looks super cool. English prompt: "A futuristic basket of fruits" (Translated: футуристическая корзина с фруктами)
Normal sampling reveals some dataset issues though, like cutting the tops off
Morning update, probably won't get much better without cleaning the dataset and some tweaks
Just finished adding the Cartoonset dataset to @huggingface 😄
Its an intermediate-level image dataset for generative modeling created by researchers at Google which features randomly generate avatar faces.
https://t.co/p3f5KZLN41
@BehemothAI makes sense, the more artistic license one takes, the harder it’s for the GAN to recognise features… guess it’s just a matter of training from much large datasets 🤷♂️
I continued adding paint to the original painting, filling in the forms and it just recognised it as a face!!!!
ok here’s the first full samples from my 100 ksteps trained model. I think these might be slightly better than MUNIT but there’s also a ton of empty images in my samples (I think bc my dataset was causing bad cropping patches)
How does Artificial Intelligence get trained?
Tune into the new podcast episode being released TOMORROW with @DatagenTech to hear more!
Datagen is a synthetic data platform that lets you design a dataset and generate it with one click.
#data #ai #artificialintelligence
@JohnnyBlokchain yep!!! working to train my own image dataset... its proving challenging as i do not have a dedicated GPU -- but here were some results. You can kinda see the skulls trying to be formed
@cirruscreek I'm sure they will come out with a version that searches the internet soon enough, the field is accelerating exponentially - new tools becoming available every week. The datasets contain millions of images. Yep I have posted many Starry ai and Wombo results