DataComp

Filtering track

Mixing track

400M-1x

1B-1x

1B-5x

7B-1x

7B-2x

*Denotes an external dataset that was not built from our competition pools but is still included as a useful reference point

Rank	Created	Submission	Core	Extended	MMLU	Authors	Writeup

*Denotes an external dataset that was not built from our competition pools but is still included as a useful reference point

Rank	Created	Submission	Core	Extended	MMLU	Authors	Writeup

*Denotes an external dataset that was not built from our competition pools but is still included as a useful reference point

Rank	Created	Submission	Core	Extended	MMLU	Authors	Writeup

*Denotes an external dataset that was not built from our competition pools but is still included as a useful reference point

Rank	Created	Submission	Core	Extended	MMLU	Authors	Writeup

*Denotes an external dataset that was not built from our competition pools but is still included as a useful reference point

Rank	Created	Submission	Core	Extended	MMLU	Authors	Writeup

*Denotes an external dataset that was not built from our competition pools but is still included as a useful reference point

Rank	Created	Submission	Core	Extended	MMLU	Authors	Writeup

*Denotes an external dataset that was not built from our competition pools but is still included as a useful reference point

Rank	Created	Submission	Core	Extended	MMLU	Authors	Writeup

*Denotes an external dataset that was not built from our competition pools but is still included as a useful reference point

Rank	Created	Submission	Core	Extended	MMLU	Authors	Writeup

*Denotes an external dataset that was not built from our competition pools but is still included as a useful reference point

Rank	Created	Submission	Core	Extended	MMLU	Authors	Writeup

*Denotes an external dataset that was not built from our competition pools but is still included as a useful reference point

Rank	Created	Submission	Core	Extended	MMLU	Authors	Writeup