Datasetpublic

placeholderlabs/pretrain-encyclopedic-mix

Normalized documents plus aligned Dolma-2 tokens and target masks. Size Tokens 7,321,112,013 (7.3B) Trainable tokens 7,321,112,013 (7.3B) Documents 6,498,683 Shards 58 UTF-8 bytes 27,544,569,761 Tokenizer allenai/dolma2-tokenizer@5292e5d6c0f4 documents.parquet - document_id, text, part_ends, part_trainable, must_not_split. The readable payload and the mask intent. metadata.parquet - one text-free row per document: token span, source, stratum, sizes… See the full description on the dataset page: https://huggingface.co/datasets/placeholderlabs/pretrain-encyclopedic-mix.

sourceHugging Faceupdated 13d agoView on Hugging Face

0likes67downloads

Dataset Card

Normalized documents plus aligned Dolma-2 tokens and target masks.

Size


Tokens	7,321,112,013 (7.3B)
Trainable tokens	7,321,112,013 (7.3B)
Documents	6,498,683
Shards	58
UTF-8 bytes	27,544,569,761
Tokenizer	`allenai/dolma2-tokenizer@5292e5d6c0f4`

—documents.parquet - document_id, text, part_ends, part_trainable, must_not_split. The readable payload and the mask intent.
—metadata.parquet - one text-free row per document: token span, source, stratum, sizes, provenance and the full source metadata.
—tokens.bin / offsets.bin / target-mask.bin - the training hot path. Little-endian int32 IDs, int64 document boundaries, one LSB-first mask bit per token. RSDB packs 4K or 16K sequences at training time.