|
|
--- |
|
|
license: cc-by-4.0 |
|
|
--- |
|
|
|
|
|
# Dataset Card for MarkMyWords |
|
|
|
|
|
This dataset is a part of the [MarkMyWords](https://github.com/wagner-group/MarkMyWords/) watermarking benchmark. |
|
|
|
|
|
We provide three tasks: book reports, story generation, and fake news. |
|
|
These do not have labeled outputs, as we care about the quality of generation relative to the original LLM, not in absolute terms. |
|
|
|
|
|
- **License:** CC-BY 4.0 |
|
|
|
|
|
### Dataset Sources |
|
|
|
|
|
- **Repository:** https://github.com/wagner-group/MarkMyWords/ |
|
|
- **Paper:** https://arxiv.org/abs/2312.00273 |
|
|
|
