Federation composition · live view

What's in the commons.

The AMPL-aligned federation, by source category. Token counts are drawn from public corpus documentation — Common Corpus (PleIAs), Common Pile (EleutherAI), Institutional Books (Harvard IDI). Click any segment to see details.

Aligned scale potential
2.27 T
tokens · across federation members · v0.1 baseline
0 500B 1.0T 1.5T 2.0T 2.27T
Open Culture
967B tokens
Public domain books, newspapers from cultural heritage repositories, and open projects like Wikisource and Project Gutenberg. The largest source category in the federation — primarily multilingual periodical archives and pre-1929 book corpora cleared through copyright expiration.
Note on framing. The federation comprises corpora stewarded by partner institutions (PleIAs, EleutherAI, Harvard IDI, Internet Archive). AIC does not curate these corpora. AIC adds the AMPL licensing layer, registry coverage per document, and federation coordination. Source totals are drawn from each partner's public documentation. Phase 2 contributions (language communities, GLAMs, public broadcasters) will appear as additional categories as they join.