Skip to content

Dataset · SLPL

Largest cleaned Farsi text corpus

SLPL/naab

A massive Persian text corpus (130 GB, 15 billion words) for training language models from scratch and fine-tuning.

What's inside

Contains cleaned Farsi texts totaling about 130 GB. The dataset includes 250 million paragraphs and 15 billion words (10 to 100 million records in metadata). According to the authors, this is the largest ready-to-use open corpus for the Persian language.

What you can do with it

  • Pre-training language models (LLMs) and encoders from scratch in Farsi
  • Continual pre-training to adapt multilingual models to the Persian language
  • Training tokenizers for Farsi texts
Tasks
Fill mask, Text generation
Languages
Persian
Size
10M–100M rows
Downloads this month
3.6K
Likes
45

Models trained on it

How to load it

from datasets import load_dataset

ds = load_dataset("SLPL/naab")