--- language: - en --- # Dataset Card for StoryDataset This dataset card provides an overview of the StoryDataset, which contains untokenized text from ten original stories. It is designed for training and fine-tuning language models on diverse narrative content. ## Dataset Details ### Dataset Description This dataset contains untokenized text from ten original stories. It is intended for training large language models (LLMs) on narrative text to enhance their understanding and generation of creative writing. - **Curated by:** Victor - **Funded by [optional]:** Victor - **Shared by [optional]:** Victor - **Language(s) (NLP):** English - **License:** None ### Dataset Sources [optional] - **Repository:** [vgealtach/TrainingData](https://huggingface.co/datasets/vgealtach/TrainingData) - **Paper [optional]:** None - **Demo [optional]:** None ## Uses ### Direct Use This dataset is suitable for training and fine-tuning language models on narrative text. It aims to improve models' capabilities in generating and understanding creative writing, storytelling, and dialogue. ### Out-of-Scope Use The dataset is not intended for training models on highly technical, scientific, or non-narrative text. It may not perform well for tasks requiring specialized domain knowledge or factual accuracy. ## Dataset Structure The dataset is structured as follows: - **Text Field:** Contains the untokenized narrative text of the stories. ## Dataset Creation ### Curation Rationale The dataset was created to provide a rich source of narrative text for training language models. It aims to improve models' abilities in generating and understanding diverse and creative storytelling. ### Source Data The data consists of ten original stories written by Victor. Each story is presented in its entirety, without tokenization, to preserve the narrative flow and structure. #### Data Collection and Processing Stories were collected and formatted into plain text files. The text was reviewed for consistency and completeness before being included in the dataset. #### Who are the source data producers? The stories were created by Victor, with no external data producers involved. ### Annotations [optional] The dataset does not include annotations. It provides raw narrative text for training purposes. #### Annotation process No annotations were added to this dataset. #### Who are the annotators? Not applicable, as no annotations were performed. #### Personal and Sensitive Information The dataset does not contain personal, sensitive, or private information. The text includes fictional stories and does not reveal any real-life personal data. ## Bias, Risks, and Limitations ### Bias, Risks, and Limitations The dataset consists solely of narratives written by Victor, which may introduce a bias towards certain styles or themes. The dataset may not represent a wide range of genres or cultural perspectives. ### Recommendations Users should be aware that the dataset reflects the author's style and themes, which may limit its applicability for training models intended to work across diverse narrative styles or cultural contexts. ## Citation [optional] **BibTeX:** ```bibtex @dataset{storydataset2024, author = {Victor}, title = {StoryDataset}, year = {2024}, url = {https://huggingface.co/datasets/vgealtach/TrainingData}, }