The full dataset is not published yet. A license review comes first, and the dataset will be released once it is finished; no date is promised. Until then the site shows the dataset's statistics and a sample, and lists cannot be built or downloaded.

Common English OpenList
Sections

Versions

A released version never changes; a correction comes as a new version with a change log.

The version and the pin

Version 1.0 is the first public version of the CEOL Universe. It is a projection of the pin, 10,296,606 candidates (pipeline/research/universe_v2/universe_v2.json). 1,985 of them are restricted, attested only by 2 lists whose terms forbid redistribution, and are in no version. The other 10,294,621 are shippable. The public projection holds back 274,471 of those (§6) and ships 10,020,150.

The release was built on October 25, 2026 by pipeline/research/release/build_release.py, at commit 6891bff41ef0. The build verifies every pinned layer's sha256 against its committed record before reading it, writes the files, re-reads them, and asserts the properties this card states before anything is kept. The record of the build, with every file's rows, columns, bytes, and sha256 and the assertions' counts, is pipeline/research/release/release_v1.0.json. A later version is built the same way, and its card's change log (§16) states what changed.

Layersha256Committed record
pin (data/universe_v2.parquet)adaa1232b6bc4d9ead97f5e19a005b264bb6db8db77b6374fe66cdd1d2aca9ddresearch/universe_v2/universe_v2.json, artifacts.universe_v2_parquet_sha256
provenance (data/universe_v2_provenance.parquet)7c9dc6284bf6393769eec3b26c27b8c77fa68dcf08cb5ef77148c2b7f07bec1bresearch/provenance/provenance.json, artifact.sha256
evidence (data/evidence.parquet)e4c86cc4a0caf3f01996e0abb50cedfd525ee7e08183679f154884bf7250d6eeresearch/evidence/evidence.json, artifacts.evidence_parquet_sha256
features (data/features.parquet)c578289b61d7b957bba1c91b9e65d3dbfcde3c5d74098316f6993643f240b7c7research/provenance/provenance.json, inputs.features_parquet_sha256
model_basis (data/eol_valid_model_basis.parquet)4395526a55acb7a2464fd6439984f478889bf65e419f41eeed272d168eb43ce0research/provenance/model_basis.json, artifact.sha256

Change log

  • 1.0 (October 25, 2026): the first public version. 10,020,150 rows in 7 configs, with the withholding of §6, NOTICES.md, and the canary, the license, and the attribution in each Parquet file's metadata.