task_categories:
- fill-mask
language:
- en
pretty_name: FragmentOfBookCorpus
size_categories:
- 1M<n<10M
Dataset Card for Fragment Of Bookcorpus
Dataset Description
A smaller sample of the bookcorpus dataset, Which includes around 100,000 lines of text. ^^^(In comparison to the original bookcorpus' 74.1~ Million lines of text)^^^
Dataset Summary
Modified and Uploaded to the hugggingface library as a part of a project. Essentially aiming at Open-Ended conversation data. This dataset is basically a fragment of the infamous bookcorpus dataset. Which aims to serve as a testing sample for those who may not want to download the entire bookcorpus dataset just for a small sample of it.
Languages
The text is written in the English language.
Dataset Structure
A simple ".txt" file which split each sentence into a new line. For a grand total of 100,000 lines.
Data Fields
The data was originally modified for training on Masked Language Modeling with BERT. However, It may be used for variety of other tasks that may require a similar dataset pattern.
Data Splits
Currently, The Dataset is one text file which is a split of the bigger (original) bookcorpus Dataset. Hence, There is only the train split (the one text file) available for download from this Dataset.
Dataset Creation
The Dataset was created from a part of the bookcorpus Dataset and was slightly modified in the way the sentences are organized.
Source Data
The source of the Data comes from the infamous BookCorpus Dataset, Available on HuggingFace at; "https://huggingface.co/datasets/bookcorpus"
Personal and Sensitive Information
The rights and Data itself is not owned by me directly. I have simply modified it according to my needs.
Licensing Information
All rights of the Data itself belong to the owners and those who contributed to the Dataset on- -HuggingFace over at; "https://huggingface.co/datasets/bookcorpus"