README.md

metadata

language:
  - vi
license: apache-2.0
size_categories:
  - 100K<n<1M
task_categories:
  - text-generation
dataset_info:
  features:
    - name: title
      dtype: string
    - name: sapo
      dtype: string
    - name: cate
      dtype: string
    - name: tags
      dtype: string
    - name: publish
      dtype: string
    - name: source
      dtype: string
    - name: content
      dtype: string
  splits:
    - name: train
      num_bytes: 5061759896
      num_examples: 1395443
  download_size: 2178488132
  dataset_size: 5061759896
configs:
  - config_name: default
    data_files:
      - split: train
        path: data/train-*

Extract from BinhVQ vnnews corpus, random select 10% for smaller continual pre-train Vietnamese LLM