Add dataset card

README.md

@@ -0,0 +1,35 @@
+---
+license: cc-by-nc-3.0
+task_categories:
+- image-to-text
+tags:
+- vision
+- ocr
+- trocr
+- handwriting-recognition
+- document-processing
+---
+
+# Tzefa Word OCR Dataset
+
+Training data for the Tzefa TrOCR word recognition model.
+
+## Related Datasets
+This dataset consolidates references to the training data used for the Tzefa word OCR model:
+- [Words-for-Ocr](https://huggingface.co/datasets/WARAJA/Words-for-Ocr) - Word crop images
+- [WORDS-ANNOTATIONS](https://huggingface.co/datasets/WARAJA/WORDS-ANNOTATIONS) - Annotation data
+
+## Task
+Single-word handwritten text recognition for Tzefa programming language keywords.
+
+## Vocabulary
+The model recognizes uppercase Tzefa keywords:
+- **Commands:** MAKEINTEGER, MAKESTR, MAKELIST, MAKEBOOL, PRINTINTEGER, PRINTSTR, ADD, SUBTRACT, MULTIPLY, DIVIDE, WHILETRUE, IFTRUE, ENDFUNC, etc.
+- **Number words:** ZERO, ONE, TWO, THREE ... ONEHUNDRED
+- **Comparisons:** EQUALS, BIGGER, BIGEQUALS
+- **Control flow:** BREAK, WHILETRUE, IFTRUE, ELSEIFTRUE, ELSE, ENDIF
+- **User-defined variable names** (arbitrary uppercase strings)
+
+## Related
+- [Word OCR Model (TrOCR)](https://huggingface.co/WARAJA/Tzefa-Word-OCR-TrOCR)
+- [Full Tzefa Pipeline](https://huggingface.co/spaces/WARAJA/Tzefa)