Previous
Previous
Summer of Open Science
Next
Next
We released the Common Pile v0.1, 8TB of public domain and openly licensed text, for training large language models and the Comma v0.1 models trained on it.