Task
Text Generation
Z
Zyphra/Zyda-2
Text Generation · Zyphra· 174.7K
odc-by
13 TB
task_categories:text-generationlanguage:enlicense:odc-bysize_categories:n>1Tregion:us
Zyda-2 is a 5 trillion token language modeling dataset created by collecting open and high quality datasets and combining them and cross-deduplication and model-based quality filtering. Zyda-2 comprises diverse sources of web data, highly educational content, math, code, and scientific papers.
# download instantly
mlforge datasets pull Zyphra/Zyda-2
mlforge datasets pull Zyphra/Zyda-2
Dataset details
Language
en
License
odc-by
Size
13 TB
Creator
Zyphra
Downloads
174.7K
Source
huggingface_datasets
Updated
2025-08-06
About Zyphra/Zyda-2
Zyda-2 is a 5 trillion token language modeling dataset created by collecting open and high quality datasets and combining them and cross-deduplication and model-based quality filtering. Zyda-2 comprises diverse sources of web data, highly educational content, math, code, and scientific papers.