Datasets:
Offer to help: viewer fix, leftover files, loading example, official page
Hello David, thanks for putting this together. It’s the most complete and current version of the eBible corpus I’ve found.
A few things I noticed:
Viewer: The dataset viewer doesn’t work because there are more than 1,000 columns. A version with one row per verse per translation would fix that and make filtering easier.
Leftover files: The older bible_corpus.parquet and metadata.csv are still next to the newer main.parquet and metadata.parquet. The page reports about 85,000 rows instead of 41,899, so the default loader may be combining both versions.
Loading example: It uses a placeholder name (“your-org/ebible-parallel”) instead of this dataset’s name.
Official page: The bible-nlp/biblenlp-corpus dataset still relies on a loading script, which no longer works with datasets 4.0+. Would it make sense to move this version over to the bible-nlp page so people find the working one?
I’m a data analyst with a home server and some time to give. Would any of these be helpful for me to work on? I’d also be glad to help with regular rebuilds. How do you generate updates now?