I would think that npm downloads would be a much better representation of usage compared to GitHub activity. GitHub activity is often a mirror of the sponsored work of the companies backing a project where npm downloads are a more direct comparison of actual usage.