README.md

metadata

source_datasets: kmfoda/booksum
license:
  - bsd-3-clause
train-eval-index:
  - config: pszemraj--booksum_1024
    task: summarization
    task_id: summarization
    splits:
      eval_split: test
    col_mapping:
      chapter: text
      summary_text: target
task_categories:
  - summarization
  - text2text-generation
size_categories:
  - 1K<n<10K

booksum - 1024 tokens max output

goal: limit max output length explicitly to prevent partial summaries being generated.

• notebook to create

info