babylm-seqlen/train_100M_256
Viewer
•
Updated
•
639k
•
43
Double Shuffled Data
Data shuffled at both the document-level, and again at the tokenized level
-
-
babylm-seqlen/train_100M_1024
Viewer • Updated • 160k • 46 -
babylm-seqlen/train_100M_16384
Viewer • Updated • 9.86k • 42 -
babylm-seqlen/train_100M_4096
Viewer • Updated • 39.8k • 37 -
babylm-seqlen/train_100M_512
Viewer • Updated • 319k • 34 -
babylm-seqlen/train_100M_8192
Viewer • Updated • 19.8k • 30 -
babylm-seqlen/train_100M_2048
Viewer • Updated • 79.8k • 32 -
babylm-seqlen/train_100M_128
Viewer • Updated • 1.28M • 50 -
babylm-seqlen/train_100M_64
Viewer • Updated • 2.56M • 44