Datasets:
Tasks:
Text-to-Image
Modalities:
Image
Languages:
English
Size:
1M<n<10M
ArXiv:
Libraries:
WebDataset
License:
| # Licences in this dataset | |
| Each part of a sample has its own terms. The per-image licence is in `index/train.parquet` and `index/validation.parquet` | |
| (column `license`) and in each sample's `.json` file. | |
| | Part | Terms | Text | | |
| |---|---|---| | |
| | Images (`.jpg`) from PD12M | Per image: dedicated under CC0 1.0 or marked public domain (Public Domain Mark 1.0) by the source | [CC0 1.0](https://creativecommons.org/publicdomain/zero/1.0/legalcode) · [PDM 1.0](https://creativecommons.org/publicdomain/mark/1.0/) | | |
| | Images (`.jpg`) from Megalith-CC0 | Dedicated under CC0 1.0 by the source | [CC0 1.0](https://creativecommons.org/publicdomain/zero/1.0/legalcode) | | |
| | Captions (`.txt`) and metadata from PD12M | Community Data License Agreement – Permissive 2.0 | `CDLA-Permissive-2.0.txt` | | |
| | Captions (`.txt`) and metadata from Megalith-CC0 | MIT License | `MIT-Megalith-CC0.txt` | | |
| | What Udgam Labs adds: the SD-VAE latents (`.npy`), the licence index, the per-sample `.json` files and this card | Community Data License Agreement – Permissive 2.0 | `CDLA-Permissive-2.0.txt` | | |
| If you share this data, CDLA-Permissive-2.0 asks you to make its text available with it, and MIT asks you to include its | |
| notice. CC0 and the Public Domain Mark ask for nothing, though crediting the sources is good practice. | |
| **What these terms do not cover.** CC0 and the Public Domain Mark concern copyright only. They do not clear the privacy | |
| or publicity rights of people shown in an image, trademarks, logos or insignia, or other rights some countries recognise. | |
| The Public Domain Mark is a label, not a licence: it records the source's view that a work's copyright has expired, and | |
| that can depend on the country. The sources and Udgam Labs cannot guarantee the status of any single work. | |