Phonikud/

heb-text

Clean Hebrew text

HeDC4-enhanced-v1.csv

Fetch from https://huggingface.co/datasets/HeNLP/HeDC4 Clean with IQR bounds forumula (Bounds: -376.0 <= length <= 1168.0) Aggressive clean: keep only sentences with Hebrew/basic punctuation Remove duplicate punctuation

HeDC4-enhanced-v2.csv

Transform HeDC4-enhanced-v1.csv to new CSV with id,text headers, , separator and " escape char (default)

HeDC4-enhanced-v3.csv.7z

Add diacritics, stress marks… See the full description on the dataset page: https://huggingface.co/datasets/Phonikud/heb-text.