LakoreAI/bert-mlm-experiments-en

Unified English MLM Pre-training Corpus (80M Rows) This dataset is a massive, diverse, multi-domain English text corpus explicitly engineered for pre-training and domain-adaptation of BERT-style models via Masked Language Modeling (MLM). It aggregates over 80 million rows of text, completely stripped of auxiliary metadata, labels, and identifiers to expose purely raw text strings. Dataset Details Repository ID: 8Opt/bert-mlm-experiments-en Total Rows: 80,489,226… See the full description on the dataset page: https://huggingface.co/datasets/LakoreAI/bert-mlm-experiments-en.