Skip to main content
llm.info

Dolma

Open Source
Datasets

3T-token open pretraining dataset

About

Dolma is Ai2's open dataset of roughly 3 trillion tokens from web, code, papers, and books, released with an open toolkit for curation. It underpins the fully open OLMo model family.

Details

Category
Datasets
License
ODC-By 1.0