Announcing the Common Pile and Comma v0.1
•
14
Common Pile
Team
community
AI & ML interests
None defined yet.
Articles
All resources related to Common Pile v0.1, an 8TB dataset of public domain and openly licensed text
An LLM pre-training dataset produced by filtering and deduplicating the raw text collected in the Common Pile v0.1
-
common-pile/arxiv_abstracts_filtered
Viewer • Updated • 2.5M • 385 • 5 -
common-pile/arxiv_papers_filtered
Viewer • Updated • 309k • 328 • 3 -
common-pile/biodiversity_heritage_library_filtered
Viewer • Updated • 16.5M • 462 • 1 -
common-pile/caselaw_access_project_filtered
Viewer • Updated • 5.5M • 208 • 5
All resources related to Common Pile v0.1, an 8TB dataset of public domain and openly licensed text
8TB of public domain and openly licensed text
An LLM pre-training dataset produced by filtering and deduplicating the raw text collected in the Common Pile v0.1
-
common-pile/arxiv_abstracts_filtered
Viewer • Updated • 2.5M • 385 • 5 -
common-pile/arxiv_papers_filtered
Viewer • Updated • 309k • 328 • 3 -
common-pile/biodiversity_heritage_library_filtered
Viewer • Updated • 16.5M • 462 • 1 -
common-pile/caselaw_access_project_filtered
Viewer • Updated • 5.5M • 208 • 5
A collection of artifacts related to Comma v0.1—a 7B parameter LLM trained on public domain and openly licensed text