Images actually used our open LAION 400m dataset plus 400m of their own. On compute there is more than enough compute available to open source now via LAION an

Images actually used our open LAION 400m dataset plus 400m of their own. On compute there is more than enough compute available to open source now via LAION and Eleuther AI to train these models, will just a bit of time.