Datasets:

qikp
/

multi

task_categories:
  - text-generation
size_categories:
  - 10M<n<100M
tags:
  - code
  - legal
  - finance
configs:
  - config_name: apple
    data_files:
      - split: train
        path: apple.jsonl.gz
  - config_name: code
    data_files:
      - split: train
        path: code.jsonl.gz

multi

multi is a multilingual dataset containing a substantial portion of text on Apple's website, as well as code from top GitHub repos that are licensed under 0BSD, MIT-0, and CC0-1.0.

This can be useful for various tasks, like tokenizer training.