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?
Β· Sign up or log in to comment