Two different corpus sizes are in circulation, and the paper will need to cite one.
mining/1getHFdatasets.py:5 (comment): "Successfully retrieved 8481 datasets"- The 2026-06-09 corpus figures (
ProcessedCommits.csv): 44,864 commits across 8,192 datasets
Most likely explanation is that the two counts come from different runs, since the Hugging Face dataset population moves. But "most likely" is not good enough for a number that will appear in a paper, and the difference (289 datasets, ~3.5%) is large enough to shift the ~2.2% candidate rate.
Needs someone to re-run 1getHFdatasets.py, record the date and the count, and either reconcile the two or state plainly that the corpus was frozen on a given date at a given size.
Two different corpus sizes are in circulation, and the paper will need to cite one.
mining/1getHFdatasets.py:5(comment): "Successfully retrieved 8481 datasets"ProcessedCommits.csv): 44,864 commits across 8,192 datasetsMost likely explanation is that the two counts come from different runs, since the Hugging Face dataset population moves. But "most likely" is not good enough for a number that will appear in a paper, and the difference (289 datasets, ~3.5%) is large enough to shift the ~2.2% candidate rate.
Needs someone to re-run
1getHFdatasets.py, record the date and the count, and either reconcile the two or state plainly that the corpus was frozen on a given date at a given size.