🚨 Issue: Too small
Objectively, 2 million posts is already a good amount, and since this dataset was collected from the Bluesky API without user consent, and there is no easy way to opt out, as stated in point 3 of the Bluesky Developer's Guide, the corpus of raw data that we can collect for AI training and research is much larger.
Honestly the only useful complaint here
If 2M is too small, what about 235M?
https://zenodo.org/records/11082879
Public domain, funded by the EU.
Great job! Also want to thank @informatiker for the 20M dataset
https://huggingface.co./datasets/informatiker/20-million-bluesky-posts
https://bsky.jazco.dev/stats says that there are currently 669.06 million posts. So I hope this record will be broken soon
I'm working on it :)
Jetstream still working, also got two local backup nodes running that I could switch to