language:

• en
  license: mit
  size_categories:
• 10K<n<100K
  task_categories:
• question-answering
  pretty_name: orca_mini_squad
  configs:
• config_name: default
  data_files:
  • split: train
    path: data/train-*
  • split: test
    path: data/test-*
    dataset_info:
    features:
  • name: id
    dtype: string
  • name: title
    dtype: string
  • name: context
    dtype: string
  • name: question
    dtype: string
  • name: answers
    struct:
    • name: answer_start
      dtype: int64
    • name: text
      dtype: string
      splits:
  • name: train
    num_bytes: 118261864.35315199
    num_examples: 67300
  • name: test
    num_bytes: 13140597.646848004
    num_examples: 7478
    download_size: 65276229
    dataset_size: 131402462.0

Dataset Card for "orca_mini_uncensored_squad_format_train"

Dataset Description

Mostly purely an exercise in data extraction and formatting for dataset usage, and cross-model usage of data.

Uncensored data, because when everything is sanitized for alignment, the data may be "pure" but is no longer untimately realistic.

Part of an effort to create more question-answering friendly datasets that can be used for specialized domain training on small models.

Dataset Summary

This is a "squad reformat" of an existing dataset located here: https://huggingface.co/datasets/julep-ai/orca_mini_uncensored

This could be swapped for squad format datasets for typical question-answering tasks with uncensored data from a partial pull of the mini-orca dataset here: psmathur/orca_minis_uncensored_dataset

Supported Tasks and Leaderboards

• 'question-answering'

Languages

The BCP-47 code for English as generally spoken in the United States is en-US and the BCP-47 code for English as generally spoken in the United Kingdom is en-GB. It is unknown if other varieties of English are represented in the data.

Dataset Structure

Train and Test splits included

Data Format

As in the squadv2 dataset, columns are: "id", "title", "context", "question", "answers": "text", "answer_start"