metadata

dataset_info:
  features:
    - name: title
      dtype: string
    - name: question
      dtype: string
    - name: responses
      sequence: string
    - name: url
      dtype: string
  splits:
    - name: train
      num_bytes: 919149577
      num_examples: 445895
  download_size: 552347573
  dataset_size: 919149577
configs:
  - config_name: default
    data_files:
      - split: train
        path: data/train-*
task_categories:
  - question-answering
language:
  - is
pretty_name: Hugi Korkar

Hugi Korkar

All of the conversations from www.hugi.is, as of 18th of March 2024.

The data was scraped from https://www.hugi.is/forsida/korkar/, it has not been cleaned and may contain conversations not for all audiences.

The purpose of this dataset is for conversation fine-tuning, and question answering.

Author: Sigurdur Haukur Birgisson