Adds `condition-3-zh-5k` and `condition-4-zh-5k` to `configs` and `dataset_info`, reading the same parquet files as the `phase-2-the-stack-v1-condition-{3,4}-zh-5k` configs (the corpora are unchanged across phases, so one copy of the files serves both names). The config table, 'Used by' lines and load example keep the short names, which now resolve. The earlier commit on this PR that pointed prose at the phase-2 names is superseded.
docs: name the real config for conditions 3 and 4 (phase-2 namespace), fix the load example
The card's config table, 'Used by' lines and usage example referred to condition-3-zh-5k and condition-4-zh-5k, which are not configs on this repo: the only cond-3/cond-4 configs are phase-2-the-stack-v1-condition-{3,4}-zh-5k. Name those, and extend the 2026-05-18 note to say the prefix is a filing location, not a Stack v1 provenance claim. Context: the cond-3 seed-123 adapter was trained from that config and inherited the phase-2 name on upload (language-decoded-lora PR #17 renames it). No data files change.
Reorganize: move Phase 2 data under phase-2-the-stack-v1/
Madison (Pfaff) Edgar
Unverified signature
964dcbe
- Prefix all existing config names with phase-2-the-stack-v1-
- Move all parquet files under data/phase-2-the-stack-v1/<config>/
- Add banner to README explaining the rename for downstream consumers
- Frees the short condition-* namespace for incoming Phase 3 builds
(The Stack v2, streamed with near-dedup)
Signed-off-by: Madison (Pfaff) Edgar <7844510+madiedgar@users.noreply.github.com>
docs: fix CITATION.cff url to match repo, set correct type (#10)