LLäMmlein: Compact and Competitive German-Only Language Models from Scratch
Paper
•
2411.11171
•
Published
•
8
The dataset is currently empty. Upload or create new data files. Then, you will be able to explore them in the Dataset Viewer.
This dataset is a strict subset of the LLaMmlein-Dataset which in turn is a strict subset of the RedPajama V2 dataset. Therefore, it retains all licenses from RedPajama V2.
This dataset has been deduplicated against the deu_Latn shard of the HPLT3.0 dataset.
More details in our preprint!
- Downloads last month
- 6