This dataset contains text records converted from local JSONL files. Each source
record had a text field, which is exposed as the dataset's text column.
Conversion cleanup:
HTML character references such as ', &, <, and " are decoded.
Conservative text cleanup removes transcript stage tags, common YouTube boilerplate, control/replacement characters, PDF percent-space artifacts, and raw LaTeX tabular… See the full description on the dataset page: https://huggingface.co/datasets/QuixiAI/fp_pt.
