Datasets:
Nick Padgett
commited on
Commit
•
1acb130
1
Parent(s):
25776ab
Updating README
Browse files
README.md
CHANGED
@@ -13,9 +13,11 @@ tags:
|
|
13 |
![PD12M](header.jpg)
|
14 |
|
15 |
# Summary
|
16 |
-
|
17 |
|
18 |
-
[
|
|
|
|
|
19 |
|
20 |
# About
|
21 |
PD12M was built and curated with [Source.Plus](https://source.plus) with the aim of resolving many of the data quality issues that arise in web-scraped training data: the presence of copyrighted material, low quality images and captions, violent or nsfw content, PII, decaying dataset quality via broken links, etc.
|
|
|
13 |
![PD12M](header.jpg)
|
14 |
|
15 |
# Summary
|
16 |
+
At 12.4 million image-caption pairs, PD12M is the largest public domain image-text dataset to date, with sufficient size to train foundation models while minimizing copyright concerns. Through the Source.Plus platform, we also introduce novel, community-driven dataset governance mechanisms that reduce harm and support reproducibility over time.
|
17 |
|
18 |
+
[Jordan Meyer](https://linkedin.com/in/jordanmeyer) [Nicholas Padgett](https://www.linkedin.com/in/nicholas-padgett-36a921a0/) [Cullen Miller](https://www.linkedin.com/in/cullen-miller-312941290/) [Cullen Miller](https://www.linkedin.com/in/lauraexline/)
|
19 |
+
|
20 |
+
[Paper](https://arxiv.org) [Datasheet](https://huggingface.co/datasets/Spawning/PD12M/blob/main/Datasheet.pdf) [Project](https://source.plus/pd12m)
|
21 |
|
22 |
# About
|
23 |
PD12M was built and curated with [Source.Plus](https://source.plus) with the aim of resolving many of the data quality issues that arise in web-scraped training data: the presence of copyrighted material, low quality images and captions, violent or nsfw content, PII, decaying dataset quality via broken links, etc.
|