Update README.md

In this dataset, you can find images accross multiple categories of the coco, laion and dataset which have been captioned manually or forked from their existing datasets. The image has 100 images per class. It has 200 images for two classes, zebra and giraffe . We are working to introduce more images within each classes with qualityful captions plateuing at 200 images per class. This is a lightweight dataset, that can help to train small diffusion models without utlizing much resource. Moreover, it can be also used to fine tune diffusion models while preventing caatastrophic forgetting.

README.md

@@ -16,4 +16,7 @@ configs:
   data_files:
   - split: train
     path: data/train-*
----
+license: mit
+task_categories:
+- text-to-image
+---