metadata
task_categories:
- fill-mask
language:
- en
size_categories:
- 10K<n<100K
Dataset description
Dataset summmary
This dataset hase been build in order to train model on natural language understanding specified in league of legends champions. This dataset has been built by scrapping various online sources relative to champions guides and/or coaching. I then used some chunking techniques in order to generate propositions based on the retrieved online knowledge and a data augmentation technique. These techniques will be further detailed.
Languages
This dataset provides only data in english
Dataset Structure
Data Fields
- label : name of the champion that the proposition is refering too.
- text : proposition refering to the champion in label
Data Splits
- train : 16663 data points
- validation : 4166 data points
TOTAL : 20829 data points