Hi,
From the paper, the dclm-pool size contains 469B tokens and 8.2B train tokens.I just want to confirm, this data is unfiltered data and thus contains around 469B tokens, right?
· Sign up or log in to comment