Datasets:
metadata
language:
- sr
pretty_name: cc100-sr derivation by JeRTeh
license: cc-by-4.0
task_categories:
- text-generation
size_categories:
- 100M<n<1B
As the title suggests, this dataset is a derivative of the cc100-sr common-crawl based dataset for Serbian. It was deduplicated on the sentence level, and transliterated into latin alphabet.
Dataset is constituted of a single txt file containing about 660 million words that can be obtianed via:
from datasets import load_dataset
dataset = load_dataset("jerteh/cc100-sr-jerteh")
Preview:
print(dataset["train"][1096953])
{'text': 'JeRTeh – Društvo za jezičke resurse i tehnologije'}