Update README.md

README.md

@@ -77,7 +77,7 @@ A line-to-text dataset for Tibetan OCR.
 - **Features:**
   - `id`: Image file identifier
   - `label`: Text transcription
-  - `url`: Source URL of the original document
+  - `image`: Image of a line of Tibetan text
 
 - **Splits:**
   - **Train:** 601,152 samples (37.3M characters)
@@ -94,7 +94,6 @@ A line-to-text dataset for Tibetan OCR.
 
 ### Out-of-Scope Use
 - Not be suitable for handwritten Tibetan texts
-- May not suitably represent contemporary digital Tibetan fonts
 
 ## Dataset Creation
 
@@ -115,7 +114,7 @@ dataset = load_dataset("openpecha/OCR-Google_Books", split="train")
 print(dataset[0])
 # {'id': 'I1KG1163750042_0025',
 #'label':'ཡིན་པས་ཆབ་སྲིད་དང་འབྲེལ་བ་བྱུང་བ་ཙམ་ལ་ངོ་མཚར་དགོས་དོན་གང་',
-# 'url': 'https://s3.amazonaws.com/monlam.ai.ocr/OCR/training_images/I1KG1163750042_0025.jpg'}
+#'image': <PIL.JpegImagePlugin.JpegImageFile image mode=RGB size=860x45>}
 ```
 
 ## Dataset Contact