---
license: unknown
viewer: false
task_categories:
- text-generation
---

TEKGEN is a training corpus utilized by Google for their [KELM](https://huggingface.co/datasets/google-research-datasets/kelm) paper. In this [repository](https://huggingface.co/datasets/google-research-datasets/kelm), I have uploaded the knowledge graphs and the original sentences used to create these graphs for the KELM dataset. Note that this dataset is not the KELM dataset itself but rather the training corpus used for the REALM model.