metadata
configs:
- config_name: fw_edu
data_files:
- split: train
path: fw_edu/*chunk.*.jsonl
- config_name: fas_Arab
data_files:
- split: train
path: fas_Arab/*.chunk.*.jsonl
- split: validation
path: fas_Arab/*.val.jsonl
- config_name: ita_Latn
data_files:
- split: train
path: ita_Latn/*.chunk.*.jsonl
- split: validation
path: ita_Latn/*.val.jsonl
- config_name: tur_Latn
data_files:
- split: train
path: tur_Latn/*.chunk.*.jsonl
- split: validation
path: tur_Latn/*.val.jsonl
- config_name: cmn_Hani
data_files:
- split: train
path: cmn_Hani/*.chunk.*.jsonl
- split: validation
path: cmn_Hani/*.val.jsonl
- config_name: arb_Arab
data_files:
- split: train
path: arb_Arab/*.chunk.*.jsonl
- split: validation
path: arb_Arab/*.val.jsonl
- config_name: ces_Latn
data_files:
- split: train
path: ces_Latn/*.chunk.*.jsonl
- split: validation
path: ces_Latn/*.val.jsonl
- config_name: dan_Latn
data_files:
- split: train
path: dan_Latn/*.chunk.*.jsonl
- split: validation
path: dan_Latn/*.val.jsonl
- config_name: deu_Latn
data_files:
- split: train
path: deu_Latn/*.chunk.*.jsonl
- split: validation
path: deu_Latn/*.val.jsonl
- config_name: ell_Grek
data_files:
- split: train
path: ell_Grek/*.chunk.*.jsonl
- split: validation
path: ell_Grek/*.val.jsonl
- config_name: fra_Latn
data_files:
- split: train
path: fra_Latn/*.chunk.*.jsonl
- split: validation
path: fra_Latn/*.val.jsonl
- config_name: hun_Latn
data_files:
- split: train
path: hun_Latn/*.chunk.*.jsonl
- split: validation
path: hun_Latn/*.val.jsonl
- config_name: ind_Latn
data_files:
- split: train
path: ind_Latn/*.chunk.*.jsonl
- split: validation
path: ind_Latn/*.val.jsonl
- config_name: jpn_Jpan
data_files:
- split: train
path: jpn_Jpan/*.chunk.*.jsonl
- split: validation
path: jpn_Jpan/*.val.jsonl
- config_name: nld_Latn
data_files:
- split: train
path: nld_Latn/*.chunk.*.jsonl
- split: validation
path: nld_Latn/*.val.jsonl
- config_name: pol_Latn
data_files:
- split: train
path: pol_Latn/*.chunk.*.jsonl
- split: validation
path: pol_Latn/*.val.jsonl
- config_name: por_Latn
data_files:
- split: train
path: por_Latn/*.chunk.*.jsonl
- split: validation
path: por_Latn/*.val.jsonl
- config_name: rus_Cryl
data_files:
- split: train
path: rus_Cryl/*.chunk.*.jsonl
- split: validation
path: rus_Cryl/*.val.jsonl
- config_name: spa_Latn
data_files:
- split: train
path: spa_Latn/*.chunk.*.jsonl
- split: validation
path: spa_Latn/*.val.jsonl
- config_name: swe_Latn
data_files:
- split: train
path: swe_Latn/*.chunk.*.jsonl
- split: validation
path: swe_Latn/*.val.jsonl
- config_name: vie_Latn
data_files:
- split: train
path: vie_Latn/*.chunk.*.jsonl
- split: validation
path: vie_Latn/*.val.jsonl
language:
- ar
- fa
- tr
- cs
- da
- de
- el
- fr
- hu
- id
- ja
- nl
- pl
- pt
- ru
- es
- sv
- vi
- zh
- it
- en
Derived from FineWeb-Edu and FineWeb-2-HQ, this data is preshuffled and split into 8 chunks (max 50GB) suitable for pre-training using Meta's Lingua Framework.