Dataset Viewer
Auto-converted to Parquet Duplicate
bos_token_id
int64
bytes_per_token_file
string
created_utc
string
dtype
string
eos_token_id
int64
holdout
dict
name
string
notes
string
num_tokens
int64
pad_token_id
int64
revision
null
shards
list
source
string
split
string
tokenizer_hash
string
unk_token_id
int64
version
int64
vocab_size
int64
1
train-bytes.bin
2026-11-14T12:34:47.295881+00:00
uint16
2
{ "doc_id_field": "id", "method": "blake2b-64", "rate": 0.005, "salt": "dolma-v1.7-2026-11-13" }
dolma_v1_7
EOS appended per document; byte lengths are original UTF-8 bytes
3,010,860,688
0
null
[ { "num_tokens": 250000000, "path": "train-00000.bin", "sha256": "36d74977406a89c16bfe5d24980a55677dfb53c1200f4b4ec01714602cbf3c3e" }, { "num_tokens": 250000000, "path": "train-00001.bin", "sha256": "2f77fa3bb2085dd0954bd5897c764f862e49a7723d18309b61433be31a5af2e7" }, { "num_token...
jsonl:sample/*.jsonl
train
cc4eaf8f386423dc512abb92c8041d589f62d1c63552a2955e250ed0b97fae2e
3
1
32,000

No dataset card yet

Downloads last month
38