Task
Text Generation
DCLM-pro is refined from DCLM using the ProX refining framework. It contains about 500B high quality tokens, ready for general language model pre-training.
DCLM-pro is refined from DCLM using the ProX refining framework. It contains about 500B high quality tokens, ready for general language model pre-training.