Update README.md


The SSL-HWD dataset provides large-scale, diverse handwriting samples across multiple challenges, including multicoloured text, blur, background interference, distortions, and highlights, reflecting real-world degradations. Compared to existing datasets (Table 1), it is the largest, with over 10M word-level instances from 852 writers. Its vocabulary is notably richer, covering alphabetic, numeric, and rare words for stronger generalization. SSL- HWD includes both labelled data for recognition and writer identification, and unlabeled data for self-supervised, semi-supervised, and domain-adaptation tasks.

README.md

@@ -1,3 +1,7 @@
----
-license: apache-2.0
----
+---
+license: apache-2.0
+language:
+- en
+size_categories:
+- 100B<n<1T
+---