Shuffled: Yes (Seed 42)
Format: Parquet files containing input_ids only.
Split Size: 100k rows per file.
HuggingFaceFW/finetranslations-edu (tur_Latn & tuk_Latn)
Ba2han/dataset_repo (Wiki Cleaned)
mlx-community/recycling_the_web-400K… See the full description on the dataset page:
https://huggingface.co/datasets/Ba2han/tokenized-1202-2.