|
--- |
|
title: README |
|
emoji: 🏆 |
|
colorFrom: blue |
|
colorTo: red |
|
sdk: static |
|
pinned: false |
|
license: mit |
|
short_description: Master Open Sourced AI, We make Datasets and rarely a model. |
|
--- |
|
|
|
# Master Open Sourced AI |
|
|
|
This is a Group of people who make Ai stuff... Mainly datasets or Stuff like TTI (text to image) models. |
|
|
|
We care about Both Open and uncensored AI, since Censoring AI, despite it's benefites for getting rid of bad content can make AIs less straight forwards. |
|
|
|
Basically this is only for posting good datasets I've made and so I can show off the models which are the best I've made. |
|
|
|
## Current WIPs: |
|
Digital Art dataset: 100-150GB with 30000-50000~ Images. |
|
|
|
## ABOUT OUR PROJECTS: |
|
|
|
If the project has WIP it is not done. |
|
|
|
### Digital Art Dataset (WIP): |
|
The dataset will be easy and will be hopefully somewhere near 100-150GB, which since an image is about a couple MB I'd say usually 3, The dataset should be at least: 30 000~ Images and at most: 50 000~ |
|
It'll just take a while since 150 GB is a lot of images and 50K is quite a sum, The dataset will also be mainly scimmed with a filter to grab out bad images. |
|
The dataset should have about half furry (because those are easy to get because of E621) images and half Anime (mostlikely from E621 but if it happens that I need more, theres always other places.) images, Sorry realistic image fans. |
|
Yes the dataset will have NSFW Images, because 1, I'm to lazy to pick them out and 2 I think it lets the model gain more anatomy knowledge. Also I'm not sure about licincing at the moment, |
|
Since there is user generated content. |
