License:
-
Corpus index: 148 datasets, 5,909 h
-
zenodo queue: 998 candidates, 681 active-task
-
Corpus index: 142 datasets, 5,876 h
-
eeglab: read event onsets from the sample field too; HBN markers were being dropped
-
publish: batch datasets into fuller commits; the Hub allows 128 commits an hour and we were exceeding it
-
log what a barren archive actually held, so a gap in the reader is visible instead of looking like an empty record
-
Corpus index: 141 datasets, 5,869 h
-
publish: delete only the files just verified; removing the folder took recordings written during the upload
-
Corpus index: 141 datasets, 5,872 h
-
carry source_archive through provenance so datasets sort into their archive folder
-
Corpus index: 141 datasets, 5,871 h
-
hbn: dedupe republished recordings by ETag; add Video-TP and the numbered aliases to the task map
-
publish: label every archive folder; an unknown one was failing the upload instead of the label
-
Corpus index: 138 datasets, 5,874 h
-
scan_corrupt: remember verified files; the publish loop was re-decoding the whole staging backlog every round
-
hbn: one folder per task kind instead of 45,000 files in one directory
-
read EEGLAB structures published under the .mat name; whole records were converting to nothing
-
hbn: shard subjects across workers, per-worker staging
-
fix: hours read a header key that never existed, so every productive unit was journalled as a failure and re-downloaded each round
-
Corpus index: 132 datasets, 5,868 h
-
HBN ingest (4,569 subjects of active-task EEG); route folders by stated source_archive
-
repo ingest: biggest archives first, ignore tiny files in give-up rule; OSF queue
-
Corpus index: 105 datasets, 5,779 h
-
publish SUMMARY.md with the corpus; three-strikes rule for failing MOABB subjects
-
Corpus index: 104 datasets, 5,735 h
-
batched toolchain: atomic writer, content tiers, repo ingests, discovery queues
-
update tools/ingest/repo_ingest.py
-
update tools/tools/enrich_repos.py
-
update tools/tools/content_tiers.py
-
update tools/tools/publish_root.py
-
update tools/tools/summarize.py
-
update tools/tools/discover_zenodo.py
-
update tools/ingest/dryad_ingest.py
-
update tools/ingest/zenodo_ingest.py
-
update tools/ingest/webdata.py
-
update tools/tests/test_format.py
-
update tools/tools/scan_corrupt.py
-
update tools/eegx/format.py
-
update tools/tools/discover_repos.py
-
update tools/tools/discover_zenodo.py
-
update tools/tools/publish_hf.py
-
update tools/ingest/moabb_ingest.py
-
moabb: keep MNE_DATA root on purge; skip PhysionetMI duplicate
-
Corpus index: 86 datasets, 5,504 h
-
Corpus index: 86 datasets, 5,504 h
-
Corpus index: 86 datasets, 5,504 h
