first commits

This dataset cleaned and dowloaded version of following dataset: https://huggingface.co/datasets/visheratin/laion-coco-nllb
The main purpose was to extract Turkish captions.

README.md

@@ -26,4 +26,15 @@ configs:
   data_files:
   - split: train
     path: data/train-*
----
+license: mit
+task_categories:
+- translation
+- text-to-image
+- image-to-text
+- image-to-image
+language:
+- tr
+- en
+size_categories:
+- 100K<n<1M
+---