README.md

metadata

dataset_info:
  features:
    - name: text
      dtype: string
    - name: summary_original
      dtype: string
    - name: summary
      dtype: string
    - name: sha256
      dtype: string
    - name: domain_prediction
      dtype: string
  splits:
    - name: train
      num_bytes: 273000417
      num_examples: 4671
  download_size: 141143180
  dataset_size: 273000417
configs:
  - config_name: default
    data_files:
      - split: train
        path: data/train-*
license: apache-2.0
task_categories:
  - summarization
size_categories:
  - 1K<n<10K

comparisons side by side of summaries from two models

• summary_original is from mixtral 8x7
• summary is from llama-3.1 70b