Permissivly licensed datasets for LLM pre-training.
-
The Common Pile v0.1: An 8TB Dataset of Public Domain and Openly Licensed Text
Paper • 2506.05209 • Published • 66 -
HuggingFaceFW/finepdfs-edu
Viewer • Updated • 49.5M • 16.6k • 98 -
PleIAs/SYNTH
Viewer • Updated • 68M • 14.2k • 277 -
allenai/dolma3_pool
Viewer • Updated • 28.9B • 28.2k • 41