{ "dataset": "uonlp/CulturaX", "subset": "tr", "use_small": true, "total_documents": 1471979, "train_samples": 1398380, "val_samples": 73599, "train_ratio": 0.95, "seed": 2357, "train_file": "/mnt/2tb_ssd/ismAIl/Model_Architecture/data/train.txt", "val_file": "/mnt/2tb_ssd/ismAIl/Model_Architecture/data/val.txt" }