priyammaz/WrenPretraining100B download history
priyammaz/WrenPretraining100B is a text generation dataset on the Hugging Face Hub. In the last 30 days it was downloaded 639 times (77 in the last 7 days), and 1,896 times in total. It ranks #27,225 among datasets by monthly downloads.
Wren Pretraining Mix (~116.14B tokens) Pre-tokenized, ready-to-train pretraining data for the Wren model — so you can skip the prep pipeline entirely and start training. Tokens are packed as Megatron-style indexed datasets (.bin = concatenated uint16 token ids, .idx = per-document off