metadata
configs:
- config_name: bkai_newssapo_title_body
data_files:
- split: train
path: bkai_newssapo_title_body/train-*
- config_name: bkai_newssapo_title_sapo
data_files:
- split: train
path: bkai_newssapo_title_sapo/train-*
- config_name: contrastive-pretraining-vi
data_files:
- split: train
path: contrastive-pretraining-vi/train-*.parquet
- config_name: crosslingual_law
data_files:
- split: train
path: crosslingual_law/train-*.parquet
- config_name: dbpedia_vn_title_text
data_files:
- split: train
path: dbpedia_vn_title_text/train-*.parquet
- config_name: finewiki-vi
data_files:
- split: train
path: finewiki-vi/train-*.parquet
- config_name: medlatec_vn_hoi_dap
data_files:
- split: train
path: medlatec_vn_hoi_dap/train-*
- config_name: msmarco-vn-pretrain
data_files:
- split: train
path: msmarco-vn-pretrain/train-*.parquet
- config_name: viepaws
default: true
data_files:
- split: train
path: viepaws/train-*.parquet
- config_name: vietnam_journal
data_files:
- split: train
path: vietnam_journal/train-*.parquet
- config_name: vinmec_vn_hoi_dap
data_files:
- split: train
path: vinmec_vn_hoi_dap/train-*
- config_name: visp_paraphrase
data_files:
- split: train
path: visp_paraphrase/train-*.parquet
- config_name: vn_news_title_context
data_files:
- split: train
path: vn_news_title_context/train-*.parquet
- config_name: vn_news_title_description
data_files:
- split: train
path: vn_news_title_description/train-*.parquet
dataset_info:
- config_name: bkai_newssapo_title_body
features:
- name: query
dtype: string
- name: document
dtype: string
splits:
- name: train
num_bytes: 81787433681
num_examples: 21974741
download_size: 0
dataset_size: 81787433681
- config_name: bkai_newssapo_title_sapo
features:
- name: query
dtype: string
- name: document
dtype: string
splits:
- name: train
num_bytes: 7163667892
num_examples: 21434465
download_size: 4329329419
dataset_size: 7163667892
- config_name: contrastive-pretraining-vi
features:
- name: query
dtype: string
- name: document
dtype: string
splits:
- name: train
num_examples: 884499
- config_name: crosslingual_law
features:
- name: query
dtype: string
- name: document
dtype: string
splits:
- name: train
num_examples: 521697
- config_name: dbpedia_vn_title_text
features:
- name: query
dtype: string
- name: document
dtype: string
splits:
- name: train
num_examples: 4632996
- config_name: finewiki-vi
features:
- name: query
dtype: string
- name: document
dtype: string
splits:
- name: train
num_examples: 1279087
- config_name: medlatec_vn_hoi_dap
features:
- name: query
dtype: string
- name: document
dtype: string
- name: url
dtype: string
- name: item_id
dtype: string
- name: specialty
dtype: string
- name: doctor_name
dtype: string
- name: date_asked
dtype: string
- name: date_answered
dtype: string
splits:
- name: train
num_bytes: 13870177
num_examples: 5994
download_size: 5164021
dataset_size: 13870177
- config_name: msmarco-vn-pretrain
features:
- name: index
dtype: int64
- name: query
dtype: string
- name: document
dtype: string
splits:
- name: train
num_examples: 5115817
- config_name: viepaws
features:
- name: query
dtype: string
- name: document
dtype: string
splits:
- name: train
num_examples: 9966
- config_name: vietnam_journal
features:
- name: query
dtype: string
- name: document
dtype: string
splits:
- name: train
num_examples: 63426
- config_name: vinmec_vn_hoi_dap
features:
- name: query
dtype: string
- name: document
dtype: string
- name: url
dtype: string
- name: category
dtype: string
- name: category_name
dtype: string
- name: doctor_info
dtype: string
- name: tags
dtype: string
splits:
- name: train
num_bytes: 36943133
num_examples: 17253
download_size: 12630230
dataset_size: 36943133
- config_name: visp_paraphrase
features:
- name: query
dtype: string
- name: document
dtype: string
splits:
- name: train
num_examples: 797352
- config_name: vn_news_title_context
features:
- name: query
dtype: string
- name: document
dtype: string
splits:
- name: train
num_bytes: 11131694228
num_examples: 2323628
download_size: 5108584169
dataset_size: 11131694228
- config_name: vn_news_title_description
features:
- name: query
dtype: string
- name: document
dtype: string
splits:
- name: train
num_examples: 2317639
| Subset | Rows |
|---|---|
bkai_newssapo_title_body |
21,974,741 |
bkai_newssapo_title_sapo |
21,434,465 |
contrastive-pretraining-vi |
884,499 |
crosslingual_law |
521,697 |
dbpedia_vn_title_text |
4,632,996 |
finewiki-vi |
1,279,087 |
medlatec_vn_hoi_dap |
5,994 |
msmarco-vn-pretrain |
5,115,817 |
viepaws |
9,966 |
vietnam_journal |
63,426 |
vinmec_vn_hoi_dap |
17,253 |
visp_paraphrase |
797,352 |
vn_news_title_context |
2,323,628 |
vn_news_title_description |
2,317,639 |
| Total | 61,378,560 |
Nguồn dữ liệu (Data Sources)
Dưới đây là danh sách các tập dữ liệu gốc (nguồn) đã được sử dụng và xử lý lại trong repository này:
| Subset | Nguồn gốc / Original Source | Ghi chú |
|---|---|---|
bkai_newssapo_title_body / _sapo |
bkai-foundation-models/BKAINewsCorpus | |
contrastive-pretraining-vi |
orionweller/contrastive-pretraining | |
crosslingual_law |
bkai-foundation-models/crosslingual | |
dbpedia_vn_title_text |
GreenNode/dbpedia-vn | DBPedia Vietnamese |
finewiki-vi |
HuggingFaceFW/finewiki | |
medlatec_vn_hoi_dap |
Medlatec.vn | Crawled Q&A data |
msmarco-vn-pretrain |
GreenNode/msmarco-vn | Translated MS-MARCO |
viepaws |
ngwgsang/ViePAWS | |
vietnam_journal |
Dữ liệu Crawl (Web) | |
vinmec_vn_hoi_dap |
Vinmec.com | Crawled Q&A data |
visp_paraphrase |
ngwgsang/ViSP | |
vn_news_title_context / _cleaned / _description |
tmnam20/Vietnamese-News | Subset all |